Systems and methods for energy cost (EC) optimization based on one or more beamforming actions in a network

Abstract

The present disclosure discloses a system (106) and a method (400) for energy cost (EC) optimization based on one or more beamforming actions in a network (108) The method may include initialization by defining input and output parameters (baseline energy cost, optimized energy cost). The method may include performing real-time data collection, systematically gathering real-time metrics and calculating the baseline energy cost. The method may include conducting AI / ML-based prediction for traffic and load forecasting and coverage demand analysis, using historical and real-time data to predict future patterns, and allowing for resource optimization and proactive adjustments. The method may include dynamically adjusting network parameters and calculating and optimizing real-time energy cost (EC), continuously monitoring the network's current EC based on dynamically adjusted parameters. Compare optimized EC with baseline EC, adjusting parameters as needed.

Description

SYSTEMS AND METHODS FOR ENERGY COST (EC) OPTIMIZATION BASED ON ONE OR MORE BEAMFORMING ACTIONS IN A NETWORKCROSS-REFERENCE TO RELATED DISCLOSURE

[0001] This application claims priority to the Indian Provisional Patent Application No. 202421059461, titled "SYSTEMS AND METHODS FOR ADAPTIVE SSB CONFIGURATION WITH INTEGRATED BEAM SHAPING, ENERGY CONSERVATION, AND OFFLOADING", filed with the Indian Patent Office on August 06, 2024.RESERVATION OF RIGHTS

[0002] A portion of the disclosure of this patent document contains material, which is subject to intellectual property rights such as, but are not limited to, copyright, design, trademark, Integrated Circuit (IC) layout design, and / or trade dress protection, belonging to Jio Platforms Limited (JPL) or its affiliates (hereinafter referred as owner). The owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent files or records, but otherwise reserves all rights whatsoever. All rights to such intellectual property are fully reserved by the owner.TECHNICAL FIELD

[0003] The present disclosure relates generally to the field of communication systems. More particularly, the present disclosure relates to systems and methods for energy cost (EC) optimization based on one or more beamforming actions in the network.DEFINITION

[0004] As used in the present disclosure, the following terms are generally intended to have the meaning as set forth below, except to the extent that the context in which they are used to indicate otherwise.

[0005] The expression “Synchronization Signal (SS)” used hereinafter in the specification refers to a broadcast signal transmitted by a base station to enable user equipment (UE) to achieve time and frequency synchronization and identify the presence and identity of a cell for initial access and mobility procedures.

[0006] The expression “Synchronization Signal Block (SSB)” used hereinafter in the specification refers to synchronization / physical broadcast channel (PBCH) block as thesynchronization signals and the PBCH channels are packed as a single block that always moves together.

[0007] The expression “SSB sweeping” used hereinafter in the specification refers to transmission of multiple narrow beams carrying control information in sequence over the intended cell area.

[0008] The expression “Signal to Interference plus Noise Ratio (SINR)” used hereinafter in the specification refers to a metric used in wireless communication to assess the quality of a received signal. It represents the ratio of the desired signal's power to the combined power of interference and noise.

[0009] The expression “Reference Signal Received Power (RSRP)” used hereinafter in the specification refers to a metric used to measure the signal strength of a specific reference signal transmitted by a cell tower. It represents the average power received from resource elements carrying cell-specific reference signals within a defined bandwidth.

[0010] The expression “Reference Signal Received Quality (RSRQ)” used hereinafter in the specification refers to a metric that measures the quality of the radio signal being received by a device (e.g., user equipment (UE)) from a network node (i.e., base station).

[0011] The expression “Coverage footprint” used hereinafter in the specification refers to an area needed to provide a desired coverage.

[0012] The expression “Cost function” used hereinafter in the specification refers to a measure of error between what the model predicts and the actual value. The cost function also refers to a loss function.

[0013] The expression “Energy Cost” used hereinafter in the specification refers to the total energy required to operate network components (e.g., base stations, core network, UE) for performing processes such as transmitting, receiving, processing, and maintaining connections across the network infrastructure and the UEs.

[0014] The expression “Baseline EC” used hereinafter in the specification refers to the minimum or reference level of energy usage required by the network or a specific network element (e.g., base station) to remain operational, regardless of traffic load.

[0015] The expression “Energy Saving” used hereinafter in the specification refers to a process of minimizing the electrical energy consumption of network infrastructure and UEs through dynamic resource management, hardware optimization, and intelligent control mechanisms.

[0016] The expression “Energy Cost (EC) prediction function” as used hereinafter in the specification refers to a mathematical or algorithmic model used to estimate the future energy consumption of network components (e.g., base stations, antennas, or the entire network) based on current and predicted operating conditions.

[0017] The expression “Random Access Channel (RACH)” as used hereinafter in the specification refers to a shared channel used in establishing an initial connection (Initial Access) between a user device and a network.

[0018] The expression “Cell throughput” as used hereinafter in the specification refers to the amount of data transmitted by a cell.

[0019] The expression “Signal to Noise and Interference Ratio (SS-SINR)” as used hereinafter in the specification refers to the ratio of the strength of the transmitted signal compared to the background noise arising.

[0020] The expression “Power boosting” as used hereinafter in the specification refers to increasing (boosting) the transmission power of downlink control channels against the power of the data channels.

[0021] The expression “Physical Random- Access Channel (PRACH)” as used hereinafter in the specification refers to the channel used by the UEs to request an uplink allocation from the base station.

[0022] The expression “Channel Quality Indicator (CQI)” as used hereinafter in the specification refers to a feedback mechanism used by the user equipment (UE) to inform the serving base station about the quality of the downlink channel.

[0023] The expression “Hybrid automatic repeat request (hybrid ARQ or HARQ)” as used hereinafter in the specification refers to a combination of high-rate forward error correction (FEC) and automatic repeat request (ARQ) error- control.

[0024] The expression “Block Error Rate (BLER)” as used hereinafter in the specification refers to a measurement type of quality. BLER = Number of erroneous blocks / total number of received blocks.

[0025] The expression “SSB beams” as used hereinafter in the specification refers to broadcast-type beams. The SSB beams cover larger areas. SSB beams are used to transmit broadcast or multicast services to multiple user devices simultaneously. These beams allow the base station to efficiently deliver content such as broadcasts, software updates, emergency alerts, and other mass communication services.

[0026] The expression “Traffic beams” as used hereinafter in the specification refers to transmitting data between the base station and user devices. These beams deliver user-specific traffic, such as internet data, video streams, voice calls, and other communication services. The traffic beams are optimized for high data rates, low latency, and reliable connectivity.

[0027] The expression “Coverage hole” as used hereinafter in the specification refers to a region where the received signal level of the serving cell and any other neighbor is below the levels required to maintain the service under a minimum level of quality and robust radio performance.

[0028] The expression “Transmission power” as used hereinafter in the specification refers to an output power level at which a transmitter emits radio frequency (RF) signals in the network, directly influencing the coverage area, signal quality, and interference levels.

[0029] The expression “Maximum Transmission Power” as used hereinafter in the specification refers to the highest power level at which a transmitter can operate to send signals.

[0030] The expression “Minimum Transmission Power” as used hereinafter in the specification refers to the lowest power level at which a transmitter can effectively communicate with its intended receiver.

[0031] The expression “Sleep Mode” used hereinafter in the specification refers to a power-saving state for network elements (e.g., base station) to reduce energy consumption while maintaining the ability to resume full activity when needed quickly.

[0032] The expression “Low-power mode” as used hereinafter in the specification refers to a state or operational mode of network elements or user equipment where power consumption is reduced by limiting or suspending certain functions, usually during periods of low traffic or inactivity, to save energy while maintaining essential connectivity.

[0033] The expression “Load Threshold” as used hereinafter in the specification refers to a predefined limit or boundary value that indicates the maximum acceptable or desired level of traffic, resource usage, or congestion in the network before specific actions are triggered.

[0034] The expression “Current load” as used hereinafter refers to the real-time measurement or estimate of the amount of traffic, resource usage, or user activity occurring in the network at a given moment or at time t.

[0035] The expression “Dynamic traffic predictions” as used hereinafter in the specification refers to estimations of upcoming network traffic loads and user behavior, generated by analyzing current and historical data, environmental factors, and user mobility, to optimize resource allocation and improve network performance.

[0036] The expression “Potential offloading action” as used hereinafter in the specification refers to a candidate network operation identified by network management or optimization algorithms that can transfer data traffic from one network segment to another, aiming to reduce congestion, balance load, and enhance energy efficiency.

[0037] The expression “Traffic Distribution” as used hereinafter in the specification refers to the spatial and temporal pattern of network traffic (e.g., data, voice) loads across different cells, frequency bands, and network nodes, reflecting how user demand varies and is managed within the network.

[0038] The expression “Proactive energy saving strategy” as used hereinafter in the specification refers to a planned and predictive approach that uses forecasting, traffic prediction, and intelligent control to adjust network operations (e.g., switching off base stations, scaling transmission power, or reallocating resources) in advance, aiming to minimize energy usage while maintaining service quality.

[0039] The expression “Power Consumption in Sleep Mode” as used hereinafter in the specification refers to an amount of electrical power consumed by a device in a low-power state (sleep mode) to conserve energy while still being able to wake up and resume normal operation quickly. It is typically much lower than the power consumed during active operation.

[0040] The expression “Traffic Load Threshold for Activating Sleep Mode” as used hereinafter in the specification refers to a specific level of network traffic or data load at which a device (such as a base station or access point) determines it is more efficient to enter sleep mode. This threshold helps optimize power consumption by allowing devices to save energy during periods of low activity.

[0041] The expression “Minimum SSB Beam Shaping Angle Width” as used hereinafter in the specification refers to the narrowest angle width for shaping the signal of a Synchronization Signal Block (SSB) beam in wireless communication. A narrower beam allows for more focused transmission, improving signal quality and reducing interference, but requires precise alignment with the intended receiver.

[0042] The expression “Maximum Capacity Threshold” as used hereinafter in the specification refers to the highest limit of data or signal capacity that a network or communication system can handle effectively without degradation in performance. Exceeding this threshold can lead to congestion, increased latency, or packet loss in the network.

[0043] The expression “UE Distribution” as used hereinafter refers to the spatial and / or statistical arrangement of the UEs across the network area. The UE distribution shows where and how many users are located within the network's coverage area at a given time.

[0044] The expression “Artificial Intelligence / Machine Learning (AI / ML) models” used hereinafter in the specification refers to a system used to analyze, learn from, and make decisions or predictions based on data collected from the network. The AI / ML models help the network operators automate, optimize, and predict various aspects of network performance, behavior, and user experience.

[0045] The expression “Capacity and Coverage Optimization (CCO)” used hereinafter in the specification refers to a process aimed at improving both the quality of service (QoS) and the network efficiency by maximizing coverage and capacity while minimizing interference and resource wastage.

[0046] The expression “Optimized CCO” as used hereinafter in the specification refers to the systematic adjustment and enhancement of network parameters to maximize coverage and capacity while minimizing interference, energy consumption, and user service degradation.

[0047] The expression “Beamforming Optimization” as used hereinafter refers to the process of adjusting and fine-tuning beamforming parameters (e.g., direction, power, phase, and shape) of the radio beams to maximize wireless communication performance while minimizing interference, energy usage, or other costs.

[0048] The expression “Beam Shaping” as used hereinafter refers to a process of adjusting the beam pattern’s shape (e.g., width, side lobes, elevation) to optimize signal coverage or limit interference. The beam shaping may widen or narrow the beam to maximize signal strength and avoid blocked areas.

[0049] The expression “Beam Steering” as used hereinafter refers to a process of dynamically pointing the beam in the direction of the user (or reflectors) based on location and movement.

[0050] The expression “Beam control” as used hereinafter refers to a process of managing when and how to switch beams, including beam selection, prioritization, and deactivation.

[0051] The expression “Beam management” as used hereinafter refers to a process of beam sweeping for scanning different beam directions to find the best signal path, beam measurement, beam reporting, beam refinement, and beam switching.

[0052] The expression “Coverage and capacity demand” as used hereinafter in specification refers to the level of network performance required to ensure that users within a specific geographic area have reliable signal access (coverage) and sufficient resources to support voice, video, and data services at acceptable quality levels (capacity).

[0053] The expression “Traffic Offloading” as used hereinafter in the specification refers to a process of redirection of user traffic from a congested network cell (or base station) to another network cell that can handle the traffic more efficiently.

[0054] The expression “Energy Cost (EC) Prediction” as used hereinafter refers to a process of estimating the future energy consumption or expenditure required to operate network components (e.g., base stations) under varying traffic loads, time periods, and environmental conditions.

[0055] The expression “Predefined Margin” as used hereinafter refers to a fixed buffer, tolerance or threshold value set in advance to account for uncertainty, variability, or safety in performance to ensure that the performance stays within acceptable limits even in worse-than- expected conditions.

[0056] The expression “Channel Quality” as used hereinafter refers to the condition or performance level of the communication channel between a transmitter (e.g., a base station) and a receiver (e.g., user equipment), measured in terms of signal strength, interference, noise, and data throughput.

[0057] These definitions are in addition to those expressed in the art.BACKGROUND

[0058] The following description of related art is intended to provide background information pertaining to the field of the disclosure. This section may include certain aspects of the art that may be related to various features of the present disclosure. However, it should be appreciated that this section be used only to enhance the understanding of the reader with respect to the present disclosure, and not as admissions of prior art.

[0059] In modern telecommunications, high-density networks comprising a large number of small and macro cells play a critical role in delivering high-speed, low-latency connections. These dense deployments are designed to enhance the capacity and performance of telecommunications systems. However, they also introduce significant challenges, particularly regarding energy consumption. The close spacing and large number of base stations and small cells lead to increased energy usage, as these components require continuous transmission and sophisticated coordination of radio resources.

[0060] Advanced technologies such as massive MEMO (Multiple Input Multiple Output) and millimeter-wave (mmWave) communications further contribute to the network's energy demands. Massive MIMO uses a large number of antennas at the base station to simultaneously serve multiple users, requiring considerable power for beamforming and signal processing. Similarly, mmWave technology, which operates in higher frequency bands for faster data transmission,demands high power to maintain adequate signal strength and coverage. These technologies become especially energy-intensive in scenarios requiring high data rates and extensive coverage.

[0061] While network densification offers clear benefits such as improved capacity and reduced latency, it also has notable drawbacks. Chief among these is the rise in operational energy costs and the increased potential for environmental impact. Energy consumption and management are becoming critical concerns as the telecommunications industry evolves toward more complex and dense architecture. This trend highlights the urgent need for robust, scalable energy management strategies to mitigate environmental impacts without compromising network performance.

[0062] A central challenge in densely deployed networks is the substantial energy required to ensure reliable coverage, high capacity, and uninterrupted connectivity. Efficient management of this energy consumption is vital to maintaining operational performance and long-term sustainability.

[0063] A comprehensive strategy that incorporates energy management and network optimization is essential to address these challenges. A core component of this strategy is an energy cost (EC) prediction function, which enables accurate energy consumption forecasting. This predictive capability supports adaptive, real-time energy-saving measures, helping to ensure that network densification does not result in unsustainable power usage. By integrating the EC prediction function, operators can enhance energy efficiency while maintaining high levels of network performance.

[0064] Furthermore, in conventional Radio Access Network (RAN) architectures, two critical functions (e.g., beam parameter tuning and load balancing) are often executed simultaneously. Beam tuning involves adjusting the direction, shape, or power of radio beams using techniques such as beamforming to optimize signal quality and provide focused coverage. In parallel, load balancing aims to distribute user traffic across multiple cells or nodes to avoid congestion and utilize network resources efficiently. Although both functions are essential for performance optimization, their joint execution introduces significant coordination challenges.

[0065] A key challenge stems from the interdependence between beam configuration and traffic load distribution. Adjusting a beam to better serve a specific user group can shift the traffic load, influencing which users are connected to which cells. Conversely, load balancing actions (e.g., offloading user equipment (UE) to a different node) affect how beams must be oriented to maintain adequate coverage. These interrelated actions create parameter entanglement, often leading to conflicting outcomes. For example, enhancing beam direction might unintentionally overload a neighboring cell, while offloading may weaken beam alignment and degrade coverage. As a result, the simultaneous optimization of these two functions increases control complexity and can hinder the achievement of stable, energy-efficient network operation.

[0066] Therefore, an efficient approach to energy management is needed to balance the goal of maintaining high-quality service with minimizing environmental impact.OBJECTS

[0067] Some of the objectives of the present disclosure, which at least one embodiment herein satisfies, are as follows:

[0068] An objective of the present disclosure is to provide a system and a method for energy cost (EC) optimization based on one or more beamforming actions in a network.

[0069] Another objective of the present disclosure is to perform adaptive SSB configuration with integrated beam shaping, energy conservation, and offloading.

[0070] Another objective of the present disclosure is to perform an SSB beam boosting, an SSB beam shaping, and an energy cost (EC) prediction-based energy saving with a potential offloading.

[0071] Yet another objective of the present disclosure is to predict energy cost (EC) for energy savings based on coverage and capacity optimization (CCO) with a potential neighboring cell offloading.

[0072] Yet another objective of the present disclosure is to dynamically select multiple SSB beam shapes with multiple beam width configurations and SSB beam power (SSB power boosting).

[0073] Yet another objective of the present disclosure is to dynamically update the multipleSSB beam shapes with the multiple beam width configurations and the SSB beam power per sector based on an energy cost (EC) function.

[0074] Yet another objective of the present disclosure is dynamically adjusting beamforming strategies to improve network coverage and capacity with energy-saving and network load processing.

[0075] Yet another objective of the present disclosure is to improve downlink (DL) coverage and SINR with energy constraints.

[0076] Yet another objective of the present disclosure is to integrate EC prediction with CCO actions.

[0077] Yet another objective of the present disclosure is to dynamically adjust transmission power to meet coverage and capacity needs without wasting energy.

[0078] Yet another objective of the present disclosure is to enable adaptive adjustments in coverage (by controlling transmission power and antenna tilts) and capacity (by adjusting resource allocation and scheduling).

[0079] Yet another objective of the present disclosure is to integrate the EC prediction with beamforming parameter optimization actions.

[0080] Yet another objective of the present disclosure is to implement an EC prediction algorithm, comprising CCO parameters, with artificial intelligence / machine learning assisted optimization actions.

[0081] Yet another objective of the present disclosure is to tolerate the EC prediction with respect to the potential offloading action with potential CCO beam optimization action capabilities.

[0082] Yet another objective of the present disclosure is to maintain energy efficiency through the CCO and advanced beamforming.

[0083] Yet another objective of the present disclosure is to perform power control and beamforming parameter optimization in CCO actions.

[0084] Other objectives and advantages of the present disclosure will be more apparent from the following description, which is not intended to limit the scope of the present disclosure.SUMMARY

[0085] In an exemplary embodiment, a method for energy cost (EC) optimization based on one or more beamforming actions in a network is described. The method comprises receiving, by a receiving unit, a plurality of first parameters and collecting, by a collection unit, one or more real-time metrics from the network. The method comprises calculating, by a processing unit, a baseline EC based on power consumption at time t. The method comprises performing, by the processing unit, one or more analysis types, to predict a plurality of second parameters, using one or more machine learning (ML) models based on the plurality of first parameters and the one or more collected real-time metrics. The method comprises upon performing, executing, by the processing unit, the one or more beamforming actions based on the plurality of first parameters and at least one second parameter of the plurality of second parameters. The method comprises activating, by the processing unit, a sleep mode based on the one or more performed beamforming actions. The method comprises calculating, by the processing unit, an optimized EC based on the plurality of first parameters and the plurality of second parameters. The method comprises comparing, by a determining unit, the optimized EC and the baseline EC. The method comprises upon determining that the optimized EC is not less than the baseline EC by a predefined margin, re-executing, by the processing unit, the one or more beamforming actions to adjust one or more beamforming parameters.

[0086] In some embodiments, the plurality of first parameters comprises maximum transmission power, minimum transmission power, power consumption in a sleep mode, a traffic load threshold for activating sleep mode, minimum synchronization signal block (SSB) beam shaping angle width, maximum capacity threshold and historical data for training the one or more ML models. The plurality of first parameters is provided by a network operator.

[0087] In some embodiments, the one or more real-time metrics comprise one or more network performance metrics, one or more traffic metrics and channel predictions. The one or more network performance metrics comprise a synchronization signal (SS)- signal to interference plus noise ratio (SINR), a SS-reference signal received power (SS-RSRP), a SS-reference signal received quality (SS-RSRQ). The one or more traffic metrics comprise traffic load at the time t, user equipment (UE) distribution, counters and key performance indicators (KPIs).

[0088] In some embodiments, the baseline EC is calculated as is the power consumption between a time intervalfrom the time t to time T, andis a time interval between measurements.

[0089] In some embodiments, the one or more analysis type comprises a traffic and load forecasting analysis, and a coverage demand analysis.

[0090] In some embodiments, the plurality of second parameters comprises one or more traffic patterns, a traffic load, a coverage demand, and one or more user mobility patterns.

[0091] In some embodiments, the one or more beamforming actions comprise beam shaping, beam steering, beam control and beam management.

[0092] In some embodiments, the execution of the one or more beamforming actions comprises determining, by the determining unit, a beamforming gain and a beamwidth based on a number of active antenna sub-arrays and determining, by the determining unit, a total energy consumption for the number of active antenna sub-arrays based on one or more antenna parameters. The one or more antenna parameters comprise the number of active sub-arrays, a transmit power per sub-array, a baseband processing power per active radio frequency (RF) chain, a power amplifier consumption, and a minimum beamforming gain. The method comprises determining, by the determining unit, a normalized cost function based on the beamforming gain, the total energy consumption, and the beamwidth and applying, by the processing unit, an optimization loop algorithm based on the minimum beamforming gain, the maximum transmission power, a target beamwidth, and a total number of sub-arrays.

[0093] In some embodiments, the optimization loop algorithm comprises calculating, by the processing unit, the beamforming gain, the minimum beamwidth, and the transmit power for each active sub-array. The optimization loop algorithm comprises calculating, by the processing unit, the total energy consumption for each active sub-array and estimating, by the processing unit,a cost function for each active array based on the number of active sub-arrays and the total energy consumption. The optimization loop algorithm comprises selecting, by the processing unit, a minimum cost function from each estimated cost function corresponding to the number of active sub-arrays and outputting, by the processing unit, an optimal energy-optimized beam configuration.

[0094] In some embodiments, the sleep mode activation comprises monitoring, by the processing unit, a real-time cell load and determining, by the determining unit, whether the realtime cell load is less than a load threshold. The sleep mode activation comprises upon determining that the real-time cell load is less than the load threshold, executing, by the processing unit, the sleep mode. The sleep mode execution comprises activating, by the processing unit, the sleep mode on the number of active sub-arrays and determining, by the determining unit, whether the realtime cell load is greater than the load threshold after a predefined time interval. The sleep mode execution comprises upon determining that the real-time cell load is greater than the load threshold, activating, by the processing unit, a normal mode on the number of active sub-arrays.

[0095] In some embodiments, the re-execution of the one or more beamforming actions comprises adjusting, by the processing unit, one or more beamforming parameters and the one or more ML models. The one or more beamforming parameters comprise SSB beam shaping angle width.

[0096] In another exemplary embodiment, a system for energy cost (EC) optimization based on one or more beamforming actions in a network is described. The system comprises a receiving unit configured to receive a plurality of first parameters. A collection unit is configured to collect one or more real-time metrics from the network. A processing unit is configured to calculate a baseline EC based on power consumption at time t and perform one or more analysis types, to predict a plurality of second parameters, using one or more machine learning (ML) models based on the plurality of first parameters and the one or more collected real-time metrics. Upon performing, the processing unit is configured to execute the one or more beamforming actions based on the plurality of first parameters and at least one second parameter of the plurality of second parameters and activate a sleep mode based on the one or more performed beamforming actions. The processing unit is configured to calculate an optimized EC based on the plurality of first parameters and the plurality of second parameters. A determining unit is configured to compare the optimized EC and the baseline EC. Upon determining that the optimized EC is not less than the baseline EC by a predefined margin, the processing unit is configured to re-execute the one or more beamforming actions to adjust one or more beamforming parameters.

[0097] In yet another exemplary embodiment, a computer program product comprising a non-transitory computer-readable medium comprising instructions that, when executed by one or more processors, cause the one or more processors to execute a method energy cost (EC) optimization based on one or more beamforming actions in a network is described. The method comprises receiving, by a receiving unit, a plurality of first parameters and collecting, by acollection unit, one or more real-time metrics from the network. The method comprises calculating, by a processing unit, a baseline EC based on power consumption at time t. The method comprises performing, by the processing unit, one or more analysis types, to predict a plurality of second parameters, using one or more machine learning (ML) models based on the plurality of first parameters and the one or more collected real-time metrics. The method comprises upon performing, executing, by the processing unit, the one or more beamforming actions based on the plurality of first parameters and at least one second parameter of the plurality of second parameters. The method comprises activating, by the processing unit, a sleep mode based on the one or more performed beamforming actions. The method comprises calculating, by the processing unit, an optimized EC based on the plurality of first parameters and the plurality of second parameters. The method comprises comparing, by a determining unit, the optimized EC and the baseline EC. The method comprises upon determining that the optimized EC is not less than the baseline EC by a predefined margin, re-executing, by the processing unit, the one or more beamforming actions to adjust one or more beamforming parameters.BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWING

[0098] The accompanying drawings, which are incorporated herein, and constitute a part of this disclosure, illustrate exemplary embodiments of the disclosed methods and systems in which like reference numerals refer to the same parts throughout the different drawings. Components in the drawings are not necessarily to scale, emphasis instead being placed upon clearly illustrating the principles of the present disclosure. Some drawings may indicate the components using block diagrams and may not represent the internal circuitry of each component. It will be appreciated by those skilled in the art that disclosure of such drawings includes disclosure of electrical components, electronic components or circuitry commonly used to implement such components.

[0099] FIG. 1A illustrates an exemplary network architecture describing a system energy cost (EC) optimization based on one or more beamforming actions in a network, in accordance with an embodiment of the present disclosure.

[0100] FIG. IB illustrates an exemplary block diagram of the system for energy cost (EC) optimization based on one or more beamforming actions in the network, in accordance with an embodiment of the present disclosure.

[0101] FIG. 2A illustrates an exemplary schematic diagram describing synchronization signal block (SSB) to Physical Downlink Shared Channel (PDSCH) beam mismatch, in accordance with an embodiment of the present disclosure.

[0102] FIG. 2B illustrates an exemplary schematic diagram describing SSB beam shaping, in accordance with an embodiment of the present disclosure.

[0103] FIG. 3 illustrates an exemplary flow diagram of a method for energy cost (EC) prediction based on one or more beamforming actions in the network, in accordance with an embodiment of the present disclosure.

[0104] FIG. 4 illustrates another exemplary flow diagram of a method for EC optimization based on one or more beamforming actions in the network, in accordance with an embodiment of the present disclosure.

[0105] FIG. 5 illustrates an exemplary block diagram of a computer system in which or with which embodiments of the present disclosure may be implemented.

[0106] The foregoing shall be more apparent from the following more detailed description of the disclosure.LIST OF REFERENCE NUMERALS100 A Network Architecture102 User Equipment104 Base Station106 System108 Network100B Block diagram112 Processor114 Memory116 Interface118 Receiving Unit120 Processing Unit122 Determining Unit124 Database128 Collection Unit130 Processing Engine200A Schematic Diagram202 Synchronization Signal Block (SSB) Beams204 Physical Downlink Shared Channel (PDSCH) Traffic Beams200B Schematic Diagram212-1 Cell-1212-2 Cell-2214 Coverage hole216-1 Synchronization Signal Block 1 (SSB-1)216-2 Synchronization Signal Block 2 (SSB-2)300 Flow Diagram400 Flow Diagram500 Computer System510 External Storage Device520 Bus530 Main Memory540 Read-Only Memory550 Mass Storage Device560 Communication Ports570 ProcessorDETAILED DESCRIPTION

[0107] In the following description, for the purposes of explanation, various specific details are set forth in order to provide a thorough understanding of embodiments of the present disclosure. It will be apparent, however, that embodiments of the present disclosure may be practiced without these specific details. Several features described hereafter can each be used independently of one another or with any combination of other features. An individual feature may not address any of the problems discussed above or might address only some of the problems discussed above. Some of the problems discussed above might not be fully addressed by any of the features described herein. Example embodiments of the present disclosure are described below, as illustrated in various drawings in which like reference numerals refer to the same parts throughout the different drawings.

[0108] The ensuing description provides exemplary embodiments only, and is not intended to limit the scope, applicability, or configuration of the disclosure. Rather, the ensuing description of the exemplary embodiments will provide those skilled in the art with an enabling description for implementing an exemplary embodiment. It should be understood that various changes may be made in the function and arrangement of elements without departing from the spirit and scope of the disclosure as set forth.

[0109] Specific details are given in the following description to provide a thorough understanding of the embodiments. However, it will be understood by one of ordinary skill in the art that the embodiments may be practiced without these specific details. For example, circuits, systems, networks, processes, and other components may be shown as components in block diagram form in order not to obscure the embodiments in unnecessary detail. In other instances, well-known circuits, processes, algorithms, structures, and techniques may be shown without unnecessary detail in order to avoid obscuring the embodiments.

[0110] Also, it is noted that individual embodiments may be described as a process that is depicted as a flowchart, a flow diagram, a data flow diagram, a structure diagram, or a block diagram. Although a flowchart may describe the operations as a sequential process, many of the operations can be performed in parallel or concurrently. In addition, the order of the operations may be re-arranged. A process is terminated when its operations are completed but could have additional steps not included in a figure. A process may correspond to a method, a function, a procedure, a subroutine, a subprogram, etc. When a process corresponds to a function, its termination can correspond to a return of the function to the calling function or the main function.

[0111] The word “exemplary” and / or “demonstrative” is used herein to mean serving as an example, instance, or illustration. For the avoidance of doubt, the subject matter disclosed herein is not limited by such examples. In addition, any aspect or design described herein as “exemplary” and / or “demonstrative” is not necessarily to be construed as preferred or advantageous over other aspects or designs, nor is it meant to preclude equivalent exemplary structures and techniques known to those of ordinary skill in the art. Furthermore, to the extent that the terms “includes,” “has,” “contains,” and other similar words are used in either the detailed description or the claims, such terms are intended to be inclusive like the term “comprising” as an open transition word without precluding any additional or other elements.

[0112] Reference throughout this specification to “one embodiment” or “an embodiment” or “an instance” or “one instance” means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present disclosure. Thus, the appearances of the phrases “in one embodiment” or “in an embodiment” in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0113] The terminology used herein is to describe particular embodiments only and is not intended to be limiting the disclosure. As used herein, the singular forms “a”, “an”, and “the” are intended to include the plural forms as well, unless the context indicates otherwise. It will be further understood that the terms “comprises” and / or “comprising,” when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. As used herein, the term “and / or” includes any combinations of one or more of the associated listed items. It should be noted that the terms “mobile device”, “user equipment”, “user device”, “communication device”, “device” and similar terms are used interchangeably for the purpose of describing the invention. These terms are not intended to limit the scope of the invention or imply any specific functionality or limitations on the described embodiments. The use of these terms is solely for convenience and clarity of description. The invention is not limited to any particular type of device or equipment, and it should be understood that other equivalent terms or variations thereof may be used interchangeably without departing from the scope of the invention as defined herein.

[0114] While considerable emphasis has been placed herein on the components and component parts of the preferred embodiments, it will be appreciated that many embodiments can be made and that many changes can be made in the preferred embodiments without departing from the principles of the disclosure. These and other changes in the preferred embodiment as well as other embodiments of the disclosure will be apparent to those skilled in the art from the disclosure herein, whereby it is to be distinctly understood that the foregoing descriptive matter is to be interpreted merely as illustrative of the disclosure and not as a limitation.

[0115] In modern telecommunications, high-density networks comprising a large number of small and macro cells play a critical role in delivering high-speed, low-latency connections. These dense deployments are designed to enhance the capacity and performance of telecommunications systems. However, they also introduce significant challenges, particularly regarding energy consumption. The close spacing and large number of base stations and small cells lead to increased energy usage, as these components require continuous transmission and sophisticated coordination of radio resources.

[0116] Advanced technologies such as massive MIMO (Multiple Input Multiple Output) and millimeter-wave (mmWave) communications further contribute to the network's energy demands. Massive MIMO uses a large number of antennas at the base station to simultaneously serve multiple users, requiring considerable power for beamforming and signal processing. Similarly, mmWave technology, which operates in higher frequency bands for faster data transmission, demands high power to maintain adequate signal strength and coverage. These technologies become especially energy-intensive in scenarios requiring high data rates and extensive coverage.

[0117] While network densification offers clear benefits such as improved capacity and reduced latency, it also has notable drawbacks. Chief among these is the rise in operational energy costs and the increased potential for environmental impact. Energy consumption and management are becoming critical concerns as the telecommunications industry evolves toward more complex and dense architecture. This trend highlights the urgent need for robust, scalable energy management strategies to mitigate environmental impacts without compromising network performance.

[0118] A central challenge in densely deployed networks is the substantial energy required to ensure reliable coverage, high capacity, and uninterrupted connectivity. Efficient management of this energy consumption is vital to maintaining operational performance and long-term sustainability.

[0119] A comprehensive strategy that incorporates energy management and network optimization is essential to address these challenges. A core component of this strategy is an energy cost (EC) prediction function, which enables accurate energy consumption forecasting. This predictive capability supports adaptive, real-time energy-saving measures, helping to ensure that network densification does not result in unsustainable power usage. By integrating the EC prediction function, operators can enhance energy efficiency while maintaining high levels of network performance.

[0120] Furthermore, in conventional Radio Access Network (RAN) architectures, two critical functions (e.g., beam parameter tuning and load balancing) are often executed simultaneously. Beam tuning involves adjusting the direction, shape, or power of radio beams using techniques such as beamforming to optimize signal quality and provide focused coverage. In parallel, load balancing aims to distribute user traffic across multiple cells or nodes to avoid congestion and utilize network resources efficiently. Although both functions are essential for performance optimization, their joint execution introduces significant coordination challenges.

[0121] A key challenge stems from the interdependence between beam configuration and traffic load distribution. Adjusting a beam to serve a specific user group better can shift the traffic load, influencing which users are connected to which cells. Conversely, load balancing actions (e.g., offloading user equipment (UE) to a different node) affect how beams must be oriented to maintain adequate coverage. These interrelated actions create parameter entanglement, often leading to conflicting outcomes. For example, enhancing beam direction might unintentionally overload a neighboring cell, while offloading may weaken beam alignment and degrade coverage. As a result, the simultaneous optimization of these two functions increases control complexity and can hinder the achievement of stable, energy-efficient network operation.

[0122] There is, therefore, a need for systems and methods to provide an efficient approach to address the problems of conventional energy management techniques to balance high-quality service delivery with minimizing environmental impact.

[0123] The present disclosure aims to overcome the above-mentioned and other existing problems in this field of technology by providing a system and a method for energy cost optimization based on one or more beamforming actions in the network.

[0124] The present disclosure aims to overcome the above-mentioned and other existing problems in this field of technology by providing a system and a method for adaptive SSB configurations integrated beam shaping, energy conservation, and offloading. The beam widths of the SSB beams are dynamically shaped along with the SSB beam boosting (e.g., SSB power boosting). This dynamic shaping of SSB beams and the SSB beam boosting leads to improvement in the SINR and keeping the corresponding coverage on a certain desired level.

[0125] In another energy-saving technique, the present disclosure may use Synchronization Signal Block (SSB) beam shaping and boosting. SSB beam shaping allows for the adjustment of beam patterns to focus on specific areas, improving signal strength and coverage while efficiently managing radio resources. By enhancing the precision of signal targeting, this technique helps reduce energy waste and improve overall network efficiency. The integration of the energy cost prediction function with the SSB beam shaping offers a data-driven approach to forecasting energy usage. This function analyzes real-time and historical network parameters, including traffic load, user density, and environmental conditions, to predict energy consumption patterns. With accurate energy forecasts, network operators can make informed decisions about operational modes, such as dynamically adjusting transmission power, optimizing beamforming configurations, and selectively activating or deactivating small cells based on current demand. For instance, during low-traffic periods, the network can reduce power levels or put certain cells into sleep mode to conserve energy.

[0126] The present disclosure may employ mutually exclusive CCO action and offloading actions into an energy cost (EC) prediction function for a base station (e.g., gNB) in the network (e.g., 5G RAN) in a more flexible way. In the evolving landscape of 5G networks, optimizing resource and energy efficiency is a key priority. Two central strategies play a vital role in achieving this: offloading and Coverage and Capacity Optimization (CCO). While these techniques have traditionally been applied together, emerging research highlights the potential advantages of treating them as independent, complementary actions to unlock greater flexibility and efficiency.

[0127] Offloading focuses on redistributing network traffic from one cell to another, helping to balance load and reduce congestion, particularly in high-demand areas. When guided by energy-aware considerations, it can also shift traffic to more energy-efficient cells, thereby reducing overall power consumption. In contrast, CCO enhances network performance by fine- tuning parameters such as antenna tilt, transmission power, and beamforming techniques to improve both coverage and capacity.

[0128] Typically, these approaches are closely intertwined with high traffic in a given area may trigger both offloading and CCO to maintain stability and performance. For example, CCOmight expand coverage using beam steering, while offloading helps lighten the load on congested cells. However, during an Energy Cost (EC) prediction process, it may be more energy-efficient to prioritize a targeted CCO adjustment (e.g., refined SSB beam-shaping, steering, or high- resolution beam prediction) over initiating an offloading action. This shift allows for improved energy savings without compromising network performance

[0129] By exploring these strategies as distinct yet coordinated tools, operators may better adapt to dynamic network conditions, reduce energy consumption, and pave the way for more sustainable 5G deployment.

[0130] To enhance energy efficiency through CCO and advanced beamforming, advanced beamforming techniques (e.g., high-resolution beam prediction) offer a powerful means to significantly improve energy efficiency in 5G networks. By precisely directing beams toward user equipment (UEs), the network enhances signal quality while minimizing energy waste. This level of precision enables a single cell to manage its traffic load more effectively, often eliminating the need for offloading by staying within its energy budget.

[0131] Coverage and Capacity Optimization (CCO) actions, including SSB and traffic beam-shaping or steering, enable the network to dynamically adapt coverage areas and focus energy where it's needed most in high-demand zones. Through careful tuning of beam parameters, the network can achieve optimal performance and energy savings, reducing the dependency on traffic offloading and maintaining consistent service quality.

[0132] Viewing offloading and CCO as independent yet complementary strategies bring added flexibility to network management. Decoupling these actions reduces operational complexity, as operators can optimize each approach independently without managing intricate interdependencies. This simplifies decision-making, promotes faster responses to network conditions, and enables more targeted energy-saving interventions.

[0133] Mutually exclusive CCO and offloading actions enhance the accuracy of energy cost (EC) predictions. Since CCO actions can be proactively designed to optimize energy use through beam control. This provides a more stable foundation for EC forecasting. In contrast, offloading, often reactive in nature, introduces unpredictability that complicates such assessments. By treating CCO and offloading as distinct processes, the network achieves more consistent and reliable estimates of energy consumption. For example, a predicted offloading scenario could be re-evaluated in favor of a more energy-efficient beam adjustment (e.g., SSB beam-shaping, steering, or high-resolution beamforming) within the CCO framework.

[0134] The present disclosure empowers 5G networks to balance performance and sustainability more effectively, paving the way for smarter, greener network management.

[0135] The primary goal is to enhance energy savings by optimizing network performance and resource usage. The algorithm follows specific rules: offloading conditions are mutuallyexclusive to CCO actions, offloading should include CCO considerations, and CCO actions should be prioritized to manage the load before considering offloading.

[0136] The present disclosure may combine energy cost prediction with offloading strategies in one effective approach. Offloading uses advanced analytics to predict energy demands based on traffic patterns, user mobility, environmental conditions, and network configurations. This enables network operators to optimize network elements, adjust transmission power, and refine beamforming techniques. For example, during periods of low demand, users can be shifted from high-power macro cells to more energy-efficient small cells. Conversely, during peak usage times, users can be directed to high-capacity cells to ensure service quality while managing energy consumption effectively.

[0137] Hereinafter, exemplary embodiments of the present disclosure will be described with reference to the accompanying drawings.

[0138] FIG. 1A illustrates an exemplary network architecture (100A) for energy cost (EC) optimization based on one or more actions in a network (108), in accordance with an embodiment of the present disclosure.

[0139] Referring to FIG. 1A, the network architecture (100A) comprises a user equipment (UE) (102), a base station (104) and a system (106). In an aspect, the system (106) may be employed in or with a digital unit. The system (106) may also be referred to as the digital unit.

[0140] In an embodiment, the user equipment (102) may include smart devices operating in a smart environment, for example, an Internet of Things (loT) system. In such an embodiment, the user equipment (102) may include, but is not limited to, smart phones, smart watches, smart sensors (e.g., mechanical, thermal, electrical, magnetic, etc.), networked appliances, networked peripheral devices, networked lighting system, communication devices, networked vehicle accessories, networked vehicular devices, smart accessories, tablets, smart television (TV), computers, smart security system, smart home system, other devices for monitoring or interacting with or for the users and / or entities, or any combination thereof. A person of ordinary skill in the art will appreciate that the user equipment (102) may include, but is not limited to, intelligent, multi-sensing, network-connected devices, that can integrate seamlessly with each other and / or with a central server or a cloud-computing system or any other device that is network-connected.

[0141] In an embodiment, the user equipment (102) may include, but is not limited to, a handheld wireless communication device (e.g., a mobile phone, a smart phone, a phablet device, and so on), a wearable computer device (e.g., a head-mounted display computer device, a headmounted camera device, a wristwatch computer device, and so on), a Global Positioning System (GPS) device, a laptop computer, a tablet computer, or another type of portable computer, a media playing device, a portable gaming system, and / or any other type of computer device with wireless communication capabilities, and the like. In an embodiment, the user equipment (102) may include, but is not limited to, any electrical, electronic, electro-mechanical, or an equipment, or acombination of one or more of the above devices such as virtual reality (VR) devices, augmented reality (AR) devices, laptop, a general-purpose computer, desktop, personal digital assistant, tablet computer, mainframe computer, or any other computing device, wherein the user equipment (102) may include one or more in-built or externally coupled accessories including, but not limited to, a visual aid device such as a camera, an audio aid, a microphone, a keyboard, and input devices for receiving input from the user or an entity such as touch pad, touch enabled screen, electronic pen, and the like. The user may be an administrator. A person of ordinary skill in the art will appreciate that the user equipment (102) may not be restricted to the mentioned devices and various other devices may be used. A person of ordinary skill in the art will appreciate that the terms “computing device(s)” and “user equipment” may be used interchangeably throughout the disclosure.

[0142] In an embodiment, the base station (104) may be a network infrastructure that provides wireless access to one or more terminals associated therewith. The base station (104) may have coverage defined by a predetermined geographic area based on the distance over which a signal may be transmitted. The base station (104) may include, but not be limited to, wireless access point, evolved NodeB (eNodeB), 5G node or next generation NodeB (gNB), wireless point, transmission / reception point (TRP), and the like.

[0143] In an embodiment, the system (106) (i.e., digital unit) may be part of the base station (104). A person of ordinary skill in the art will appreciate that the terms “system” and “digital unit” may be used interchangeably throughout the disclosure. The digital unit may be a baseband unit. In an embodiment, the digital unit may include, but not be limited to, transceivers, baseband unit (BBU), remote radio unit (RRU), radio network control units, and one or more processors associated thereto. The system (106) / digital unit may perform various operations using software algorithms (e.g., AL / ML algorithms).

[0144] In an embodiment, the network (108) may include, by way of example but not limitation, at least a portion of one or more networks having one or more nodes that transmit, receive, forward, generate, buffer, store, route, switch, process, or a combination thereof, etc. one or more messages, packets, signals, waves, voltage or current levels, some combination thereof, or so forth. The network (108) may also include, by way of example but not limitation, one or more of a wireless network, a wired network, an internet, an intranet, a public network, a private network, a packet-switched network, a circuit-switched network, an ad hoc network, an infrastructure network, a Public-Switched Telephone Network (PSTN), a cable network, a cellular network, a satellite network, a fiber optic network, or some combination thereof.

[0145] As shown in FIG. 1 A, a plurality of SSB beams between the base station (104) and the user equipment (UE) (102). The UE (102) uses a physical random-access channel (PRACH) to send a connection request to the base station (104). The base station (104) sends an acknowledgment to the connection request. After establishing connection between the base station (104) and the UE (102), the base station (104) receives of a plurality of signals and measurements 1(e.g., Hybrid Automatic Repeat Request (HARQ), Channel Quality Indicator (CQI), etc.) from the UE (102).

[0146] The base station (104) in wireless communication is a critical infrastructure component that serves as the central point of communication for mobile devices within a specific area. The base station (104) facilitates the transmission and reception of signals to and from the UE 102, such as smartphones, tablets, and loT devices, enabling them to connect to the wider network.

[0147] In an embodiment, the system (106) is configured to perform energy cost (EC) optimization based on one or more beamforming actions (as explained in detail in FIG. IB). The system (106) is configured to use software algorithms (e.g., AI / ML algorithms) to select and adjust input parameters. For example, the input parameters may include maximum transmission power, minimum transmission power, power consumption in sleep mode, a traffic load threshold for activating sleep mode, a minimum SSB beam shaping angle width, and a maximum capacity threshold. The system (106) is configured to select the maximum transmission power and calculate a baseline EC.

[0148] In an aspect, for performing one or more beamforming actions, the system (106) is configured to select an appropriate beam shape, an appropriate SSB power for each SSB beam of the plurality of SSB beams using AI / ML algorithms for desired coverage and SINR and to calculate an optimized EC. The plurality of SSB beams is part of a beam-sweeping set per sector. In this way, the plurality of SSB beam shapes has the plurality of SSB beam width configurations with SSB beam power. The selection of the appropriate beam shape, the appropriate SSB power helps to minimize the intra-cell and inter-cell beam overlapping. Also, preserving the sector coverage and maximizing the SINR for inter-site distance and geography. The selective multi SSB configurations and beam shaping improve the DL coverage and throughput. Antenna beams radiation pattern in azimuth (0) and vertical (<p), is estimated along with the side lobes. Further, the projection to ground level is performed with simple geometry, considering the tilting angle and the SSB beam tilting characteristics. Further, an optimized EC is calculated based on the input parameters and the predicted traffic load. The optimized EC is compared with the baseline EC to enable EC predication recalibration when one action renders the other suboptimal.

[0149] FIG. IB illustrates an exemplary block diagram (100B) of the system for energy management in the network (108), in accordance with an embodiment of the present disclosure.

[0150] The system (106) comprises a processor (112), a memory (114), an interface (116), a database (124) and a processing engine (130). The processing engine (130) comprises a receiving unit (118), a processing unit (120), a determining unit (122), and a collection unit (128).

[0151] In an aspect, the system (106) may include one or more processor(s) (112). The one or more processor(s) (112) may be implemented as one or more microprocessors, microcomputers, microcontrollers, edge or fog microcontrollers, digital signal processors, central processing units,logic circuitries, and / or any devices that process data based on operational instructions. The processor (112) is configured to fetch and execute computer-readable instructions stored in a memory (114) of the system (106). Among other capabilities, one or more processor(s) (112) may be configured to fetch and execute computer-readable instructions stored in the memory (114) of the system (106). The memory (114) may be configured to store one or more computer-readable instructions or routines in a non-transitory computer-readable storage medium, which may be fetched and executed to create or share data packets over a network service. The memory (114) may comprise any non-transitory storage device including, for example, volatile memory such as Random- Access Memory (RAM), or non-volatile memory such as Erasable Programmable Read- Only Memory (EPROM), flash memory, and the like. The memory (114) is configured to store one or more computer-readable instructions or routines in a non-transitory computer-readable storage medium, fetched and executed to create or share data packets over a network service.

[0152] In an embodiment, the system (106) may include an interface(s) (116). The interface(s) (116) may include a variety of interfaces, for example, interfaces for data input and output devices, referred to as I / O devices, storage devices, and the like. The interface(s) (116) may facilitate communication of the system (106). The interface(s) (116) may also provide a communication pathway for one or more components of the system (106). Examples of such components include, but are not limited to, processing unit / engine(s) (120, 130) and a database (124). The interface (116) is configured to provide a communication pathway for one or more components of the system (106).

[0153] The processing engine(s) (130) may be implemented as a combination of hardware and programming (for example, programmable instructions) to implement one or more functionalities of the processing engine(s) (130). In the examples described herein, such combinations of hardware and programming may be implemented in several different ways. For example, the programming for the processing engine(s) (130) may be processor-executable instructions stored on a non-transitory machine-readable storage medium, and the hardware for the processing engine(s) (130) may comprise a processing resource (for example, one or more processors), to execute such instructions. In the present examples, the machine-readable storage medium may store instructions that, when executed by the processing resource, implement the processing engine(s) (130). In such examples, the system (106) may comprise the machine- readable storage medium storing the instructions and the processing resource to execute the instructions, or the machine-readable storage medium may be separate but accessible to the system (106) and the processing resource. In other examples, the processing engine(s) (130) may be implemented by an electronic circuitry.

[0154] In an embodiment, the processing engine (130) may be an artificial intelligence (Al) / a machine learning (ML) engine. The AI / ML engine may automate a process of performing adaptive SSB configuration with integrated beam shaping, energy conservation, and offloading by employing one or more Al models (used interchangeably with the term AI / ML models). In some embodiments, the AI / ML engine may include one or more pre-trained AI / ML models that may beconfigured to perform dynamic shaping of plurality of SSB beams with the SSB beam boosting (e.g., SSB power boosting). The AI / ML models may provide feedback on the results of monitoring results in the system (106) for closed-loop actions and for continuous optimization and configuration of one or more parameters. In some embodiments, the one or more parameters may include, but are not limited to, network latency jitter, SINR, RSRP, uplink random access channel (UL RACH) counter, CQI reports, HARQ Key Performance Indicators (KPIs), Block Error Rate (BLER), etc.

[0155] In an embodiment, the database (124) may comprise data that may be either stored or generated as a result of functionalities implemented by any of the components of the processor(s) (112) or the processing engine(s) (130) or the system / digital unit (106).

[0156] The components of the processing engine (130) may be configured to energy cost (EC) optimization based on one or more beamforming actions in the network. Further, the components of the processing engine (130) are configured to perform a plurality of operations for adaptive SSB configuration with integrated beam shaping, energy conservation, and offloading.

[0157] The receiving unit (118) is configured to receive a plurality of first parameters. The plurality of first parameters comprises maximum transmission power, minimum transmission power, power consumption in a sleep mode, a traffic load threshold for activating sleep mode, minimum synchronization signal block (SSB) beam shaping angle width, maximum capacity threshold and historical data for training the one or more ML models. The plurality of first parameters is provided by a network operator.

[0158] The collection unit (128) is configured to collect one or more real-time metrics from the network. The one or more real-time metrics comprise one or more network performance metrics, one or more traffic metrics and channel predictions. The one or more network performance metrics comprise a synchronization signal (SS)- signal to interference plus noise ratio (SINR), a SS-reference signal received power (SS-RSRP), a SS-reference signal received quality (SS-RSRQ). The one or more traffic metrics comprise traffic load at the time t, user equipment (UE) distribution, counters and key performance indicators (KPIs).

[0159] In an aspect, the traffic metrics are vital for monitoring and optimizing the network performance. The network operators use the traffic metrics to understand how the network is being used at a given time (e.g., at time t) and how efficiently the network supports user demands. The traffic load at the time t refers to the amount of data or signaling activity being carried by any network element (e.g., cell, sector, or base station) at the time t. The traffic load reflects the utilization level of network resources (e.g., bandwidth, time slots, or radio channels). The traffic load comprises, but is not limited to, data traffic, signaling traffic and voice traffic.

[0160] In an aspect, the UE distribution refers to how the UEs are geographically and logically spread across the network at the time t. The UE distribution provides insights, for example, where users are located, how many are connected to each cell, and how this distributionchanges over time due to mobility or usage patterns. The UE distribution indicates the number of the UEs per cell / sector (i.e., how many active or idle UEs are connected to each base station. For example, Cell A has 400 connected UEs and Cell B has 90 connected UEs. The UE distribution includes spatial distribution, indicating where the UEs are located geographically (e.g., indoors, near highways, in dense urban areas). The spatial distribution helps to identify hotspots or coverage holes. The UE distribution includes temporal variation that indicates how the UE distribution changes by time of day (e.g., rush hour vs. nighttime). Furthermore, the mobility patterns include tracking how UEs move between cells, which are used for optimizing handovers and predicting congestion. The UE distribution includes traffic per UE and how much traffic each UE generates (e.g., heavy vs. light users).

[0161] In an aspect, the counters are quantitative metrics automatically collected by one or more network elements (e.g., base stations, core nodes, etc.) to monitor performance, detect issues, and support optimization. The counters are used to monitor, but are not limited to, connection requests, setup failures, utilization, handover attempts, success rate, setup success, paging requests, disconnections, packet loss rate, call drop, latency, streaming failures, etc.

[0162] In an aspect, the KPIs are quantitative metrics used to evaluate the performance, reliability and quality of service (QoS) of the network. The KPIs comprise, but are not limited to, cell setup success, call drop, handover success, radio resource utilization, signal strength, signal distribution, throughput, latency, connection setup, session establishment, paging, link availability, etc.

[0163] In an aspect, the collection unit (128) is configured to collect information about the channel prediction based on physical obstructions. The processing unit (120) may collect information about channel modeling predictions. In examples, the processing unit (120) may use Al and / or ML models to predict channel conditions based on physical obstructions (e.g., building, trees, terrain, wall, hills, etc.).

[0164] The processing unit (120) is configured to calculate a baseline EC. The baseline EC refers to the minimum or fixed amount of energy consumed by the system or network node (e.g., base station) even when the system or device is idle or operating at minimal load. The baseline EC represents the constant overhead energy usage required to keep the system (e.g., base station) powered on and ready to operate, regardless of the traffic or usage level.

[0165] The baseline EC is calculated based on power consumption at time t as given in Equation 1Equation 1Where, PCUrrent(t) is the power consumption between a time interval from the time t to time T, and t is a time interval between measurements. In an aspect, the time t represents current time.

[0166] Equation 1 describes total power consumed by the system (e.g., base station) over time period t=0 to t=T. For example, let’s consider Δt = 1 hour, T = 3 hours and power usage at each hour Pcurrent(0) =500 W, Pcurrent(0) =700 W, Pcurrent(0) =8500 W and Pcurrent(0) =600 W. So, Etotai = (500+700+800+600) = 2600Wh =2.6 kWh.

[0167] The processing unit (120) is configured to perform one or more analysis types to predict a plurality of second parameters, using one or more machine learning (ML) models based on the plurality of first parameters and the one or more collected real-time metrics. The one or more analysis types comprise a traffic and load forecasting analysis, and a coverage demand analysis.

[0168] In an aspect, the traffic and load forecasting analysis refers to an analysis to predict network usage patterns to proactively manage resources, optimize performance, and improve energy efficiency. The traffic and load forecasting analysis is used to predict future values of traffic load (data volume in Mbps / GB), number of active users (UEs), resource utilization (e.g., physical resource blocks (PRBs), spectrum, channels) over time at specific cells, base stations, or regions. For example, to forecast the next 6 hours of downlink traffic for a cluster of urban LIE cells, input data (e.g., last 48 hours of traffic data, time-of-day, and weather) and training data are provided to the AI / ML model. The AI / ML model provides output (i.e., 6 future traffic load values per cell) based on the input data and the training data. In an aspect, the traffic patterns refer to the recurring behavior or trend in how network traffic (e.g., data, voice, or signaling) is generated, distributed, and consumed over time and space. For example, in a business district, traffic is low at night, peaks between 9 AM and 5 PM, and drops after work hours. In residential areas, traffic peaks in the evening as users stream video or browse the internet. The traffic load refers to the amount of network resource usage (e.g., bandwidth, time, or frequency) by the users or applications in a specific cell or base station over a given time period. For example, to predict the downlink traffic load (in Mbps) for Cell lOl from 6 PM to 12 AM for Friday, the AI / ML model uses historical data (e.g., traffic every hour for the last 30 days) to forecast the traffic load from 6 PM to 12 AM for Friday, the predicted traffic load = 18:00 = 120 Mbps, 19:00 = 145 Mbps, , 20:00 = 160 Mbps, , 21:00 = 140 Mbps, 22:00 = 100 Mbps, and 23:00 =75 Mbps.

[0169] In an aspect, the coverage demand analysis refers to an analysis to identify where, when, and how much wireless coverage is needed based on user behavior, traffic, and network conditions. The coverage demand represents the geographical areas and time windows where users require cellular connectivity, current signal strength or capacity may be insufficient or expansion, optimization, or reconfiguration is needed. For example, the coverage demand analysis to identify areas that currently lack sufficient coverage but are likely to have high demand in the next 24 hours. The coverage demand analysis identifies cell ID 101 of city mall area, having predicted high coverage demand and current coverage quality = poor (RSRP < -110 dBm) and cell ID 310 of stadium, having predicted very high coverage demand due to the upcoming match and current coverage quality = low.

[0170] Upon performing the analysis types, the processing unit (120) is configured to execute the one or more beamforming actions based on the plurality of first parameters and at least one second parameter of the plurality of second parameters. The one or more beamforming actions comprise beam shaping, beam steering, beam control and beam management. In an aspect, the beam shaping adjusts the beam pattern’s shape (e.g., width, side lobes, elevation) to optimize signal coverage or limit interference. Physical obstruction modeling helps determine where to widen or narrow the beam to maximize signal strength and avoid blocked areas. In an aspect, the beam steering dynamically points the beam in the direction of the user (or reflectors) based on location and movement. The channel prediction enables user movement, allowing beams to preemptively shift toward likely positions (e.g., in fast mobility or non-line of sight (NLoS) conditions). In an aspect, the beam control oversees when and how to switch beams, including beam selection, prioritization, and deactivation. The predictive channel models inform which beams are likely to remain viable, reducing frequent switching and improving stability. In an aspect, the beam management comprises beam sweeping, measurement, reporting, recovery, and refinement procedures. Physical modeling improves beam selection probability, filtering of irrelevant beams, and recovery decisions when a beam drops due to blockage. Table 1 describes the beamforming actions.Table 1

[0171] The execution of the one or more beamforming actions comprises the determining unit (122) configured to determine a beamforming gain and a beam width based on a number of active antenna sub-arrays. In an aspect, the antenna refers to a physical component that radiates or receives electromagnetic signals. The antennas convert electrical signals into radio waves (for transmission) or radio waves into electrical signals (for reception). The sub-arrays refer to a group of antenna elements within the antenna (i.e., antenna divided into multiple sub-arrays).

[0172] In an aspect, the beamforming gain refers to an improvement in signal quality (e.g., signal-to-noise ratio (SNR)) achieved by focusing radio waves on a specific direction using beamforming technology. The beamforming gain (G) depends on the number of active antenna sub-arrays. The beamforming gain is given in Equation 2Where, Nactive= Number of active antenna sub-arrays and GBF = Beamforming gain

[0173] Equation 2 describes that the beamforming gain increases with the number of active antenna sub-arrays. As the number of active antenna sub-arrays increases, the beam becomes narrower and concentrates more energy in a specific direction, which enhances the beamforming gain. This results in higher signal power received by the receiving device.

[0174] In an aspect, the beamwidth is the angular width of the main lobe of an antenna’s radiation pattern. The beamwidth (0) is inversely proportional to the aperture. Aperture refers to the effective size of the antenna array and is proportional to the number of antenna elements and their spacing. The beamwidth is given in Equation 3Where, Θ = Beamwidth, λ = wavelength, and Daperture= Effective aperture size,

[0175] Equation 3 describes as the aperture increases with an increase in number of active antenna sub-arrays or a larger physical array, the beam becomes narrower. Conversely, a smaller aperture results in a wider beam. A larger aperture enables better directional performance by producing narrower beams and higher gain. This allows the antenna to resolve signals from more precise directions, improving spatial selectivity. Narrow beamwidth (i.e., small 0) is desirable in beamforming, as it focuses energy more effectively and reduces interference from unwanted directions.

[0176] The determining unit (122) is configured to determine total energy consumption for the number of active antenna sub-arrays based on one or more antenna parameters. The one or more antenna parameters comprise the number of active sub-arrays, a transmit power per subarray, a baseband processing power per active radio frequency (RF) chain, a power amplifier consumption, and a minimum beamforming gain. In an aspect, the RF chain refers to a hardware path (comprising amplifiers, converters, filters, oscillators, mixers, etc.) that processes signals between a digital baseband and the antenna.

[0177] In an operative aspect, total energy consumption is calculated as given in Equation 4Where, Nactiveis number of active antenna sub-arrays (or RF chains), Ptxis transmit power per subarray, baseband processing power per active RF chain (Pbb), and ηpais the amplifier efficiency (e.g., 1 / 0.4). In an aspect, the baseband processing power (Pbb) refers to power used for digital signal processing, encoding, modulation, etc. The transmit power (Ptx) refers to actual power sentover the air through antenna sub-arrays. The amplifier efficiency refers to a measure of how effectively an amplifier converts input power into output power.

[0178] Equation 4 describes total energy consumption as the sum of power consumption of all active antenna sub-arrays. The total energy consumption increases as the number of active antenna sub-arrays increases. For example, Nactive= 4, Pbb= 1W, Ptx= 0.5W, ηpa=1 / 0.4=2.5. So,E( Nactive, Ptx) = 4*(l+0.5*2.5) = 4*2.25= 9 W.

[0179] The determining unit (122) is configured to determine a normalized cost function based on the beamforming gain, the total energy consumption, and the beamwidth. In an aspect, a cost function in the network refers to a mathematical expression that quantifies cost, penalty, or loss associated with a particular network configuration, resource allocation, or operational decision. The cost function is used to evaluate and optimize network performance by assigning a numerical value to how well a solution meets desired objectives such as minimizing energy consumption, maximizing throughput, reducing interference, or balancing load. In an aspect, the normalized cost function is a mathematical tool used to evaluate and optimize network performance by combining multiple performance metrics (e.g., energy consumption, throughput, latency, beamforming gain, or bandwidth) after scaling them to a common range (e.g., between 0 and 1). Lower normalized cost (i.e., close to 0) is better, indicating high beamforming gain, low energy consumption, and high bandwidth. Higher normalized cost (i.e., close to 1) indicates poor performance.

[0180] In an operative aspect, the normalized cost function is calculated as given in Equation 5Equation 5Where, a,P,y are tuneable weights, E(Nactive, Ptx) is actual energy consumption, ( Greq—GBF(Nactive))2is a penalty if gain is below target, and is a penalty for deviating from desired beamwidth.

[0181] Equation 5 describes minimizing the cost function by optimally selecting the number of active antenna sub-arrays and the transmit power per sub-array, such that the energy efficiency, beamforming performance (gain), and spatial resolution (angular precision) are effectively balanced.

[0182] The processing unit (120) is configured to apply an optimization loop algorithm based on the minimum beamforming gain, the maximum transmission power, a target beamwidth, and a total number of sub-arrays. In an aspect, the optimization loop algorithm refers to an iterativecomputational process used to improve the network performance step-by-step by repeatedly evaluating and adjusting variables (beamforming gain, transmission power, beamforming angle and width), and refining the AI / ML models until reaching desired output.

[0183] In the optimization loop algorithm, the processing unit (120) is configured to calculate the beamforming gain, the minimum beamwidth, and the transmit power for each active antenna sub-array. In an operative aspect, number of active antenna sub-arrays (Nactive) = 1 to M. For each active antenna sub-array, the beamforming gain is calculated as given in Equation 6GBF = 10 • log 10(Nactive) Equation 6

[0184] Equation 6 describes that as the number of active antenna sub-arrays increases; the beamforming gain increases correspondingly.

[0185] For each active antenna sub-array, the minimum beam width θ( Nactive) is calculated in Equation 7.

[0186] Equation 7 describes that the beamwidth improves approximately proportional to the inverse square root of the number of active antenna sub-arrays. Further, increasing the number of active antenna sub-arrays narrows the beam, enhancing spatial resolution.

[0187] For each active antenna sub-array, the transmit power Ptxis calculated as given in Equation 8.Ptx= min{Pmax / Nactive, required per-user power} Equation 8

[0188] Equation 8 describes that the transmit power is limited either by maximum transmission power evenly divided among the number of active antenna sub-arrays or by power required per user.

[0189] The processing unit (120) is configured to calculate the total energy consumption for each active sub-array. In an operative aspect, the total energy consumption for each active antenna sub-array is calculated based on the number of active antenna sub-arrays (Nactive), baseband processing power per active radio frequency (RF) chain (Pbb), transmit power per antenna sub-array (Ptx) and amplifier efficiency ( ηpa). The total energy consumption is given in Equation 9Etotal=Nactive* ( Pbb+ ηpa* Ptx) Equation 9

[0190] Equation 9 describes total energy consumption as the total power used by the number of active antenna sub-arrays, including baseband processing and amplifier power consumption.

[0191] The processing unit (120) is configured to estimate a cost function for each active array based on the number of active sub-arrays and the total energy consumption. In an aspect, the cost function for each active array (C(Nactive, Ptx)) is evaluated to check each antenna array configuration and select the one antenna array that minimizes the total energy consumption by considering constraints, for example, the beamforming gain, the total transmit power, and the minimum beamwidth. The constraint corresponding to the beamforming gainwhere GBF(Nactive) is the beamforming gain of the number of active sub-arrays and Greq is the minimum required beamforming gain. GBF(Nactive) is greater than or equal to the required gain Greq. Condition 1 i.e.,ensures that the system supports the current load without dropping below the minimum service quality. Further, the constraint corresponding to the total transmit power, Nactive’Ptxis total transmit power for the number of active subarrays, and Pmax is maximum power. The total transmit power (i.e., transmit power per subarray *number of active sub-arrays) is less than or equal to the maximum power. Condition 2 i.e.,ensures that total power consumed by the active sub-arrays stays within the system’s maximum power allowed limit. The constraint corresponding to minimum beamwidthθ(Nactive) is total beamwidth for the number of active sub-arrays, and θmax is maximum beamwidth. The total beamwidth for the number of active sub-arrays is less than or equal to the maximum beamwidth. Conditionensures that the beamwidth stays within maximum beamwidth for system performance, security, and hardware capabilities.

[0192] The processing unit (120) is configured to select a minimum cost function from each estimated cost function corresponding to the number of active sub-arrays. In an aspect, one on more antenna sub-arrays having minimum cost function is selected from estimated cost function of each antenna sub-array. The minimum cost function is given in Equation 10.C(Nactive, Ptx) = argmin C Equation 10

[0193] Equation 10 describes the values of the number of active antenna sub-arrays and the transmit power are those values that minimize the cost function. Minimizing the cost function is essential to balance performance with energy efficiency, hardware constraints, and operational costs, thereby ensuring that the system operates optimally without wasting resources or violating constraints.

[0194] The processing unit (120) is configured to output an optimal energy-optimized beam configuration. In an aspect, beam configuration of antenna sub-array having C(Nactive, Ptx) = argmin C is an optimal energy-optimized beam configuration i.e., optimal (Nactive, Ptx). The beam configuration of antenna sub-array having C(Nactive, Ptx) = argmin C is output as the optimal energy-optimized beam configuration. The optimal energy-optimized beam configuration is applied on the number of active sub-arrays (Nactive).

[0195] The processing unit (120) is configured to activate a sleep mode based on the one or more performed beamforming actions. In the sleep mode activation, the processing unit (120)is configured to monitor a real-time cell load. The determining unit (122) is configured to determine whether the real-time cell load is less than a load threshold. Upon determining that the real-time cell load is less than the load threshold, the processing unit (120) is configured to execute the sleep mode. In an aspect, after performing the beamforming actions, real-time cell load is again monitored to check the sleep mode activation condition (i.e., LCurrent< Lth). Upon detecting that real-time cell load is less than the load threshold, the sleep mode is activated. In the sleep mode, current transmission power is equal to transmission power in the sleep mode i.e., PCurrent(t) = Psleep.

[0196] In the sleep mode execution, the processing unit (120) is configured to activate the sleep mode on the number of active sub-arrays. In an operative aspect, the sleep mode is activated on Nsleepsub-arrays by setting Nactive= Nsleep. The Nsleepis minimum number of sub-arrays to form a low-power wide beam.

[0197] In the sleep mode, the beamforming gain is reduced to a beamforming gain corresponding to the sleep mode by setting the beamforming gain as given in Equation 11

[0198] Equation 11 describes that the beamforming gain depends on the number of antenna sub-arrays set in sleep mode. The broad coverage is maintained by setting beam to wide beam with 9 sleep-

[0199] Further, in the sleep mode, the transmit power is set as given in Equation 12Ptx= min(P max / Nsleep, required power) Equation 12

[0200] Equation 12 describes that the transmit power in the sleep mode is minimum due to either maximum transmission power by number of antenna sub-arrays in sleep mode or required power per antenna sub-array.

[0201] The total energy consumption in the sleep mode is calculated as given in Equation 13Etotal sleep= Nsleep* (Pbb+ ηpa* Ptx) Equation 13

[0202] Equation 13 describes total energy consumption in the sleep mode is the power required by number of antenna sub-arrays in the sleep mode, baseband processing power and transmit power per antenna sub-array.

[0203] Further, low-energy wide beam is applied to the Nsleepsub-arrays.

[0204] The determining unit (122) is configured to determine whether the real-time cell load is greater than the load threshold after a predefined time interval. In an aspect, after the activation of the sleep mode, the real-time cell load (LCurrent) is continuously monitored to resumea normal mode. After the predefined time interval (e.g., 5 min), the real-time cell load is compared with the load threshold. In an aspect, the predefined time interval is a fixed, predetermined duration set, used to schedule a trigger to check a condition that the real-time cell load (LCurrent) is greater than the load threshold (Lth). The predefined time interval is configured by the network operator based on network requirements, for example, usage traffic, and performance metrics.

[0205] Upon determining that the real-time cell load is greater than the load threshold, the processing unit (120) is configured to activate a normal mode on the number of active sub-arrays. In an aspect, upon determining that the real-time cell load is greater than the load threshold (LCurrent> Lth), the normal mode is activated on the number of active sub-arrays i.e., Nactive= 1 to M. Further, full precision beams are resumed when the real-time cell load exceeds the load threshold.

[0206] Furthermore, after activating the number of active sub-arrays (i.e., Nactive= 1 to M), the beamforming gain (GBF) and the beamwidth (0) are computed for each active sub-array (Nactive).

[0207] The beamforming gain (GBF) is calculated as GBF = 10 1ogl 0( Nactive). The beamwidth is calculated as Further, the minimum transmit power (Ptx) iscalculated based on required user quality of service (QoS). The total energy consumption is calculated as E = Nactive* (Pbb+ ηpa* Ptx).

[0208] The cost function is computed as given in Equation 14Equation 14

[0209] The cost function of the sub-array satisfying the condition C(Nactive*, Ptx*) = argmin C is selected. The beam configuration of the sub-array of the selected cost function is considered as an optimal beam configuration. The optimal beam configuration is applied on the number of active sub-arrays (Nactive).

[0210] The processing unit (120) is configured to calculate an optimized EC based on the plurality of first parameters and the plurality of second parameters. In an aspect, the optimized EC is calculated as minimax optimized function. In an aspect, the minimax optimization function is very useful in our analysis to find a solution that minimizes the maximum value of a function, often within a given set of constraints. The minimax optimization function is used in scenarios where there is a conflict or a worst-case scenario to consider. A value for one variable that minimizes the maximum value of another variable is determined, and the second variable is chosen to maximize the function. In an operative aspect, the minimax optimization function is used when the base station (i.e., gNB) wants to minimize its maximum possible power consumption.

[0211] In an aspect, the optimized EC is calculated based on the maximum transmission power, the minimum transmission power and the predicted load at the time t. The optimized EC is calculated as given in Equation 15 Equation 15

[0212] Equation 15 describes the optimized EC (ECopt) is the sum of optimal transmission power Popt(t) from time t=0 to time t=T. The optimized EC (ECopt) is calculated as the total optimized energy consumption by dynamically adjusting transmit power based on predicted load, while respecting minimum and maximum power constraints, then summing over all time intervals (from time t=0 to time t=T). In an aspect, the optimized EC is determined to minimize the total energy usage over a given period while ensuring that performance requirements (e.g., coverage, capacity, or quality of service (QoS)).

[0213] The determining unit (122) is configured to compare the optimized EC and the baseline EC. In an aspect, the optimized EC is compared with the baseline EC to validate that the optimization reduces energy consumption while meeting the constraints (i.e., beamforming gain, total transmit power and minimum beamwidth). The baseline EC serves as a reference point or benchmark that reflects current operations.

[0214] Upon determining that the optimized EC is not less than the baseline EC by a predefined margin, the processing unit (120) is configured to re-execute the one or more beamforming actions to adjust one or more beamforming parameters. In an aspect, the predefined margin refers to as a specified threshold or minimum allowable difference that quantifies the level of improvement required for the optimization to be considered significant or effective. For example, the baseline EC (before optimization) = 100 units, the predefined margin (the minimum energy saving required) = 10 units. For the optimization to be considered effective, the optimized EC < 90 units. When the optimized EC = 94 units, energy savings (i.e., 100 - 94) = 6 units. 6 units saved is less than the predefined margin of 10 units. The optimized EC is not less than the baseline EC by the predefined margin, so the optimization is not accepted. Further, the optimized EC = 88 unit, the energy savings (i.e., 100 - 88) = 12 units. 12 units saved exceed the predefined margin of 10 units. The optimization is accepted, because it achieves energy savings.

[0215] In the re-execution of the one or more beamforming actions, the processing unit (120) is configured to adjust one or more beamforming parameters and the one or more ML models. The one or more beamforming parameters comprise SSB beam shaping angle width. In an aspect, the beamforming parameters (i.e., beam-shaping, beam-steering) are adjusted to fulfill the condition ECopt«ECbaseat the time period T.

[0216] At the continuous learning and feedback loop step, various steps such as performance evaluation, model refinement, and optimization refinement are performed. In theperformance evaluation, the network performance of optimizations is evaluated in terms of energy savings and network quality. In the model refinement, the AI / ML models are updated with new data for improved prediction accuracy. In the optimization refinement, CCO strategies and parameters are refined based on performance evaluations and feedback.

[0217] Although FIG. IB shows the exemplary block diagram (100B) of the system (106), in other embodiments, the system (106) may include fewer components, different components, differently arranged components, or additional functional components than depicted in FIG. IB. Additionally, or alternatively, one or more components of the system (106) may perform functions described as being performed by one or more other components of the system (106).

[0218] As advanced features continue to enhance network performance, there is growing attention on optimizing their energy consumption. The energy demands of 5G infrastructure present an opportunity to innovate in ways that reduce operational costs and minimize environmental impact. Incorporating an energy cost prediction function alongside traditional Cell Coverage Optimization (CCO) techniques offers a proactive approach to achieving more sustainable and cost-efficient network operations.

[0219] According to another implementation, the processing unit (120) may be configured to calculate baseline energy cost (denoted by ECbase) and optimized energy cost (denoted by ECopt) based on one or more input network parameters. In examples, the input network parameters may include maximum transmission power (denoted by Pmax), minimum transmission power (denoted by Pmin), power consumption in sleep mode (denoted by Psleep), traffic load threshold for activating sleep mode (denoted by Lth), minimum SSB beam shaping angle width (denoted by B-min), maximum capacity threshold (denoted by Cmax), and historical data for training artificial intelligence (Al) and / or machine learning (ML) models. Additionally, the processing unit (120) may be configured to collect real-time data including information about real-time metrics, channel prediction, and neighboring cells data and capacity.

[0220] In examples, the processing unit (120) may collect information about real-time metrics including network performance metrics and traffic metrics. In examples, the network performance metrics may include Synchronization Signal - Signal to Interference plus Noise Ratio (SS-SINR) metrics, Synchronization Signal - Reference Signal Received Power (SS-RSRP) metrics, and Synchronization Signal - Reference Signal Received Quality (SS-RSRQ) metrics. Further, in examples, the traffic metrics include information about current traffic load (number of Radio Resource Control (RRC) connected UEs) and UE distribution.

[0221] According to an implementation, the processing unit (120) may also collect information about channel prediction based on physical obstructions. The processing unit (120) may collect information about channel modeling predictions. In examples, the processing unit (120) may use Al and / or ML models to predict channel conditions based on physical obstructions. Further, the processing unit (120) may collect information about neighboring cells’ load andcapacity. The processing unit (120) may be configured to calculate the baseline energy cost and the optimized energy cost based on the collected real-time data.

[0222] In an implementation, the processing unit (120) may calculate the baseline energy cost based on current power consumption without Coverage and Capacity Optimization (CCO) optimizations. For example, the processing unit (120) may calculate the baseline energy cost using Equation provided below.where, ECbaserepresents baseline energy cost and PCurrent(. ) represents power consumption at time t during the time interval of measurements and dt.

[0223] According to an implementation, the processing unit (120) may use Al and / or ML- based prediction for traffic and load forecasting, coverage demand analysis, and neighboring cell assessment. In an implementation, the processing unit (120) may use Al and / or ML algorithms to predict future traffic patterns and load distribution based on historical and real-time data. The processing unit (120) may also predict coverage demand and user mobility patterns.

[0224] According to an implementation, the processing unit (120) may perform dynamic adjustment of the input network parameters. In examples, the dynamic adjustment of the network parameters includes adaptive transmission power control. In an implementation, the processing unit (120) may perform adaptive transmission power control to calculate optimal transmission power. In an implementation, the processing unit (120) may calculate the optimal transmission power based on predicted traffic load and coverage requirements. In examples, the processing unit (120) may calculate the optimal transmission power using Equation provided below.where Popt represents optimal transmission power, Pminrepresents minimum transmission power, Pmaxrepresents maximum transmission power, and Lpred(t) represents the predicted traffic load at time t.

[0225] According to an implementation, the processing unit (120) may perform beamforming optimization. In an implementation, the processing unit (120) may be configured to adjust beamforming angles and widths based on user density and demand. The processing unit (120) may calculate an optimized beam using Equation provided below.where Bopt(t) represents optimized beam, Bminrepresents minimum SSB beam shaping angle width, and UEpredt) represents the predicted UE distribution.

[0226] In an implementation, the processing unit (120) may activate sleep mode for gNB. In examples, the processing unit (120) may activate the sleep mode based on criteria for sleep mode. In examples, if current load (Lcurrent) is significantly reduced post offloading, the processing unit (120) may transition the gNB to sleep mode. In example, the sleep mode for the gNB is represented using Equation provided below.

[0227] In an implementation, the processing unit (120) may perform real-time energy cost calculation and optimization. In examples, the processing unit (120) may calculate the optimized energy cost using the Equation provided below.where, ECoptrepresents optimized energy cost (EC).

[0228] In an aspect, the processing unit (120) may be further configured to compare the calculated Optimized EC with the baseline EC. Based on the comparison, the present disclosure may adjust optimization parameters if ECoptis not sufficiently lower than ECbase.

[0229] In an aspect, the processing unit (120) may be configured to employ a learning and feedback loop. In an aspect, the processing unit (120) may opt for a performance evaluation, a model refinement, and an optimization refinement for continuous learning and a feedback loop. In the performance evaluation, the present disclosure may evaluate the performance of optimizations in terms of energy savings and network quality. In the model refinement, the present disclosure may update AI / ML models with new data for improved prediction accuracy. In the optimization refinement, the present disclosure may refine CCO strategies and parameters based on performance evaluations and feedback.

[0230] Hence, there is an increasing concern about energy consumption associated with these advanced features. The high energy demands of 5G infrastructure can lead to significant operational costs and environmental impacts. Therefore, integrating an energy cost prediction function alongside traditional CCO techniques becomes crucial.

[0231] The present disclosure discloses a set of different SSB beam shapes with different beam width configurations and SSB beam power (SSB power boosting), to be dynamically selected and adjusted by the AI / ML optimization algorithm per sector, considering an algorithm for energy savings based on an Energy Cost (EC) function. Further, the present disclosure involves predicting energy consumption based on real-time network data and adjusting network operations accordingly to minimize energy use without compromising service quality. By incorporating thispredictive capability, network operators can balance maintaining optimal performance and reducing energy expenditures.

[0232] The present disclosure employs an algorithm to dynamically select, among the available beam width beams, the proper beam shape, the most suitable SSB power per individual SSB beam of the beam-sweeping set per sector, under the constraint of desired energy savings, coverage and SINR. The goal is to minimize the intra-cell and inter-cell beam overlapping, preserve the sector coverage and maximize the expected SINR for different inter-site distances and geography. Hence, ultimately, the multi-SSB beamwidth beams will improve the DL coverage and throughput.

[0233] In one aspect, the system (106) may be configured to integrate SSB beam shaping, SSB beam boosting, and energy cost prediction, enabling a more sustainable and efficient 5G network operation. This integrated approach facilitates dynamic adjustments to beamforming strategies, which can be tailored to optimize coverage and capacity and align energy-saving goals.

[0234] Incorporating an energy cost prediction function into the Coverage and Capacity Optimization (CCO) framework represents a significant advancement towards sustainable network management. One of the primary aspects of CCO is the dynamic adjustment of transmission power to meet coverage and capacity needs without wasting energy. This not only promotes responsible use of energy resources but also supports global sustainability goals. As 5G networks continue to grow and evolve, adopting such proactive energy management strategies will play a vital role in harmonizing technological progress with environmental responsibility and cost-effectiveness.

[0235] FIG. 2 A illustrates an exemplary schematic diagram (200 A) describing SSB to Physical Downlink Shared Channel (PDSCH) beam mismatch, in accordance with an embodiment of the present disclosure.

[0236] As shown in FIG. 2A, the base station (104) transmits a plurality of SSB beams (202-1, 202-2, 202-3, 202-4) towards the UE (102). A person of ordinary skill in the art will understand that one or more SSB beams (202-1, 202-2, 202-3, 202-4) may collectively be referred to as the SSB beams (202). A plurality of PDSCH traffic beams (204-1, 204-2, 204-3, 204-4) are also transmitted between the base station (104) and the UE (102). A person of ordinary skill in the art will understand that one or more PDSCH traffic beams (204-1, 204-2, 204-3, 204-4) may collectively be referred to as the PDSCH traffic beams (204).

[0237] The SSB beam is the beam through which the base station transmits synchronization signals that the UE uses to detect the cell and measure channel quality. Meanwhile, the PDSCH beam is the beam used to send the actual user data to the UE. As both types of beams are present between the base station (104) and the UE (102), the SSB to PDSCH beam mismatch occurs due to gaps between the shapes of the SSB beams (202) and the PDSCH traffic beams (204). When the SSB and PDSCH beams differ a situation (also known as SSB to PDSCH beam mismatch), the channel quality measurements based on the SSB beam do notaccurately reflect the quality of the PDSCH beam. This discrepancy leads to challenges in communication performance because the UE’s decisions and network scheduling are based on potentially misleading signal quality information.

[0238] This beam mismatch negatively impacts energy efficiency, coverage, and capacity optimization (CCO). The UE wastes power attempting to decode weaker PDSCH beams than expected, while the base station may increase transmission power or use less efficient beamforming, raising overall energy consumption. False channel quality reports cause coverage holes, poorer cell-edge performance, and suboptimal handover or expansion decisions. Mismatched beams result in inefficient scheduling, reduced spectral efficiency, increased retransmissions, and higher latency. Additionally, managing and mitigating these mismatches introduces extra signalling overhead, further reducing the usable capacity of the network.

[0239] FIG. 2B illustrates an exemplary schematic diagram (200B) describing SSB beam shaping, in accordance with an embodiment of the present disclosure.

[0240] As shown in FIG. 2B, a cell 1 (212-1) and a cell 2 (212-2) are present between a network node-1 (104-1) and a network node-2 (104-2). An SSB 1 (216-1) and an SSB 2 (216-2) are transmitted from the network node-2 (104-2) to the cell 2 (212-2). As seen in FIG. 2B, the shape of the SSB 1 (216-1) does not cover the upper edge of the cell 2 (212-2), which leads to a coverage hole (214) in the cell 2 (212-2).

[0241] To solve the problem of the coverage hole (214) in cell 2 (212-2), the beam shaping is performed on the SSB 1 (216-1). After performing the SSB 1 beam shaping, the shape of the SSB beam covers the coverage hole formed in the cell 2 (212-2). By SSB 1 beam shaping, the problem of the coverage hole (214) is solved. The shape, direction, and power of the SSB beams are adjusted through beam shaping to fill areas of the coverage hole. By shaping the SSB beams (i.e., making the beams wider, steering towards underserved areas, or increasing the power in specific directions), the base station may improve signal strength in those coverage holes. This targeted adjustment helps ensure that UEs receive strong enough synchronization signals to access and maintain connection with the network. Ultimately, the SSB beam shaping reduces coverage gaps, improves initial cell detection, and enhances overall user experience by providing more consistent and reliable coverage across the cell.

[0242] In an embodiment, a switch in the coverage area of the SSB 1 (216-1) may be taken without any effect on a coverage area of the SSB 2 (216-2). In order to obtain a coordination of coverage between the cell 1 (212-1) and the cell 2 (212-2), an indication of switching in the coverage area of the SSB 1 (216-1) and the SSB 2 (216-2) is provided to the cell 1 (212-1) and cell 2 (212-2) over an interface (e.g., Xn interface).

[0243] In an aspect, on changing the shape of the SSB beam in one sector, the coverage footprint is estimated for that sector. On detecting any error or anomaly, or the coverage footprint is found to be less than the configurable threshold (i.e., higher threshold), the shape of the SSBbeam is again changed in that sector. In this way, the SSB beam shaping is a synchronizing process that is being performed throughout the sector.

[0244] The present disclosure employs a combined algorithmic approach, based on a functional integration of several system functional integrities like SSB beam shaping, SSB beam steering, Energy Cost (EC) function and offloading techniques, which will enhance the system performance and preserve the green energy network retainability.

[0245] The Energy Cost prediction function plays a critical role in managing the energy efficiency of dense networks. The SSB multi-beam management allows for the simultaneous control and optimization of multiple beams within a network. The Offloading, or handover, actions are critical in maintaining optimal load distribution across the network. In dense deployments, efficient handover mechanisms ensure that users are seamlessly transitioned between cells, minimizing congestion and optimizing resource utilization. The Energy Cost prediction function can inform these handover decisions by predicting the energy costs associated with different handover scenarios. Finally, the SSB beam shaping involves adjusting the beam's shape to focus signal energy on specific directions, enhancing coverage and signal quality in targeted areas. Azimuth and elevation angle steering further refine this by adjusting the direction of the beams in three-dimensional space. These capabilities are particularly important in dense networks, where precise control over beam direction can significantly reduce interference and improve user experience.

[0246] The integration of these technologies and functions offers a comprehensive approach to energy savings and system performance optimization. The predictive capabilities of the energy cost prediction function, combined with the dynamic control offered by SSB multibeam management, beam shaping, and steering, allow for real-time adjustments that balance energy consumption with network performance. Offloading mechanisms complement these efforts by efficiently distributing user load, reducing the strain on any single cell and optimizing overall network efficiency.

[0247] The present disclosure is related mainly to gNB CU / DU and / or 0AM functional blocks. In an embodiment, the present disclosure addresses SSB (Synchronization Signal Block) multi-beam management in dense 5G and upcoming 6G networks. This approach integrates several key components: energy cost prediction, offloading (handover) actions, beam shaping, and beam steering (azimuth / elevation).

[0248] In an aspect, the following steps outline the system's process for implementing SSB multi-beam management while prioritizing energy savings and optimal network performance:

[0249] 1. Initialization

[0250] Input Parameters:- Network topology and cell layout- User equipment (UE) distribution and mobility patterns- Current network traffic load- Energy consumption models for network elements- Beamforming capabilities (beam width, angles)- Quality of Service (QoS) requirements

[0251] Initialization of System States:- Initial energy consumption state- Beam configuration and current orientations- Active UEs and their associated beams / cells- Predicted traffic demand and UE mobility patterns

[0252] 2. Energy Cost Prediction

[0253] Data Collection:- Monitor real-time data on traffic load, user activity, and environmental conditions.- Collect historical data to build predictive models.- Prediction Model Application:- Utilize machine learning or statistical models to forecast energy consumption based on traffic patterns, user density, and predicted load.- Estimate the energy cost for different beam configurations and network states.- Decision Making:- Calculate the potential energy savings for various configurations.- Identify configurations that balance energy efficiency with QoS requirements.

[0254] 3. Offloading (Handover) Actions

[0255] Trigger Detection:- Identify conditions that necessitate handovers, such as deteriorating signal quality, congestion, or energy-saving potential.

[0256] Handover Decision:- Determine the optimal target cell / beam for each UE based on:- Signal strength- Predicted load and energy costs- QoS requirements and user preferences

[0257] Execution:- Execute the handover, ensuring minimal disruption and maintaining connectivity.- Update system states to reflect the new UE associations and energy metrics.

[0258] 4. SSB Beam Shaping

[0259] Optimization Goals:- Maximize signal coverage and quality.- Minimize interference with adjacent beams / cells.- Align with energy cost predictions.

[0260] Beam Pattern Adjustment:- Calculate the optimal beam shape (e.g., width, power distribution) for each sector or cell based on the current UE distribution and predicted traffic.

[0261] Implementation:- Dynamically adjust the antenna array settings to achieve the desired beam shape.- Monitor the impact on network performance and energy consumption.

[0262] 5. SSB Beam Azimuth / Elevation Angle Steering

[0263] Target Determination:- Identify the optimal azimuth and elevation angles for each beam to maximize coverage and signal strength for active UEs.- Consider physical obstacles, user density, and mobility patterns.

[0264] Angle Calculation:- Calculate the desired angles based on the spatial distribution of UEs and the network's energy-saving objectives.

[0265] Adjustment Execution:- Adjust the beamforming hardware / software to steer the beams to the calculated angles.- Continuously monitor and adapt to changes in the environment and user distribution.

[0266] 6. Feedback Loop and Monitoring

[0267] Continuous Monitoring:- Real-time monitoring of network performance metrics (e.g., signal quality, user throughput, energy consumption).- Regular updates of predictive models with new data.

[0268] Feedback and Adjustment:- Use performance data to refine energy cost predictions, beam configurations, and handover strategies.- Implement automatic adjustments based on monitored data to maintain optimal performance and energy efficiency.

[0269] 7. Optimization and Learning

[0270] Machine Learning Integration:- Implement reinforcement learning or other Al techniques to continually optimize the system based on observed outcomes.

[0271] Adaptive Learning:- Allow the system to learn from past configurations and outcomes, improving prediction accuracy and decision-making over time.

[0272] 8. Conclusion and Future Adjustments

[0273] Evaluation:- Periodically evaluate the performance of the multi-beam management system.- Compare energy savings, QoS metrics, and user satisfaction against targets.

[0274] Future Planning:- Plan for future adjustments based on evolving technology, user needs, and environmental factors.

[0275] As mobile networks grow more complex with diverse traffic patterns and evolving user behaviors, traditional optimization methods become inadequate in addressing the dynamic and real-time requirements of modern networks.

[0276] AI / ML offers a robust, data-driven approach to predict and manage these conditions effectively. The disclosure describes the processes for training, validating, and deploying AI / ML models, ensuring that they meet the high standards necessary for effective CCO.

[0277] The integration of AI / ML technologies in CCO is a crucial step toward developing self-organizing networks (SON). SON can autonomously manage and optimize network parameters, such as load balancing, interference management, and the dynamic configuration of network resources. This autonomy allows networks to adapt to changing conditions and demands,thereby enhancing overall efficiency and user experience. The use of AI / ML in network management not only addresses current challenges but also sets the stage for future innovations in telecommunications.

[0278] SSB beamwidth refers to the angular width of the beam used to transmit synchronization signals. Proper management of SSB beamwidth is essential for ensuring optimal coverage and capacity. Narrower beams can provide higher gain and better coverage in specific directions, while wider beams can cover larger areas with lower gain. SSB beam steering refers to the elevation angle and / or the azimuth angle. The trade-off between azimuth beamwidth, azimuth / elevation beam steering and coverage area is crucial for network performance.

[0279] Efficient energy consumption can also be achieved by other means such as reduction of load, coverage modification, or other RAN configuration adjustments. The optimal energy saving decision depends on many factors, including the load situation at different RAN nodes, RAN nodes' capabilities, KPI / QoS requirements, number of active UEs, UE mobility, cell utilization, etc.

[0280] Potential improvements with SSB Beamwidth Management are:

[0281] Enhanced Coverage - By dynamically adjusting the SSB beamwidth based on realtime network conditions, AI / ML algorithms can ensure that coverage is maximized in areas with high user density or poor signal quality. This adaptive beamwidth management allows for more efficient utilization of network resources, reducing coverage gaps and enhancing overall user experience.

[0282] Improved Capacity - AI / ML algorithms can analyze traffic patterns and predict areas where capacity enhancements are needed. By narrowing the SSB beamwidth in these areas, the network can provide higher signal strength and better data rates to users, effectively increasing network capacity. This is particularly beneficial in urban environments with high user density.

[0283] Interference Management -Interference is a significant challenge in dense network deployments. By intelligently managing the SSB beamwidth, AI / ML algorithms can minimize interference between neighboring cells. Narrower beams can be used to reduce overlap and interference, while wider beams can be employed in less congested areas to maintain coverage without causing excessive interference.

[0284] Energy Efficiency - Adjusting SSB beamwidth can also contribute to energy efficiency. By focusing energy only where it is needed, the network can reduce unnecessary power consumption. AI / ML algorithms can optimize beamwidth settings to balance coverage and capacity needs with energy efficiency goals.

[0285] The management of SSB (Synchronization Signal Block) beamwidth, along with azimuth and elevation beam steering, offers various potential improvements for mobile networkperformance. These improvements leverage AI / ML algorithms to dynamically adapt network parameters based on real-time conditions, thus enhancing coverage, capacity, interference management, and energy efficiency. Here are potential improvements focusing on both beamwidth and beam azimuth / elevation steering:

[0286] Potential improvements with SSB Beamwidth Management and beam steering are:

[0287] Enhanced Coverage- Dynamic Beamwidth Adjustment: By adjusting the beamwidth dynamically, networks can maximize coverage in areas with high user density or poor signal quality. For instance, wider beamwidths can cover larger areas, which is useful in sparsely populated regions.- Adaptive Beam Steering: Azimuth and elevation steering can be used to direct beams more precisely towards areas with higher demand, ensuring better signal penetration and reducing dead zones. This is especially beneficial in complex terrains or urban environments with high-rise buildings.

[0288] Improved Capacity- Targeted Narrow Beams: Narrowing the beamwidth in areas with high traffic demand can increase signal strength and data rates, thereby boosting network capacity. This is crucial in urban hotspots or events with dense user concentrations.- Optimal Beam Steering Angles: By optimizing the azimuth and elevation angles, the network can focus beams more effectively on clusters of users, enhancing data throughput and reducing signal interference with other users.

[0289] Interference Management- Interference Mitigation Through Beamwidth Control: In dense network deployments, narrower beams can be utilized to reduce overlap and interference between adjacent cells. This minimizes cross-cell interference and improves the quality of service.- Directional Steering to Avoid Interference: By steering beams away from interference-prone areas or towards underutilized spectrum segments, networks can further mitigate interference issues. This is especially useful in areas with dense infrastructure or high- frequency reuse.

[0290] Energy Efficiency- Energy Conservation with Focused Beamwidth: By narrowing the beamwidth and targeting specific areas, the network can reduce power wastage, focusing energy only where it is needed. This leads to a more efficient use of power resources.- Steering for Optimal Power Usage: Adjusting the azimuth and elevation angles can also contribute to energy efficiency. For example, in less congested areas, beams can be steeredto broader coverage with lower power settings, while in high-demand areas, more precise, high-power beams can be used.- Training AI / ML models for CCO involves using vast amounts of network data, including user mobility patterns, signal strength measurements, traffic loads, and environmental factors. This data is used to train algorithms to recognize patterns and predict network performance under varying conditions.

[0291] The training process includes several stages:

[0292] Data Collection and Preprocessing: Collecting data from various network elements and preprocessing it to ensure quality and relevance.

[0293] Feature Selection: Identifying key features that influence network coverage and capacity.

[0294] Model Training: Using supervised or unsupervised learning techniques to train the model. Techniques such as neural networks, decision trees, or support vector machines can be used depending on the complexity of the task.

[0295] Validation and Testing: Validating the model with a separate dataset to ensure accuracy and generalizability. This step includes testing the model under different scenarios to evaluate its performance.

[0296] Typically, once trained, the AI / ML model is deployed in the network to perform real-time inference. The model continuously analyzes incoming data to make predictions and recommendations for optimizing coverage and capacity.

[0297] Key aspects of the inference process include:

[0298] Real-Time Data Processing: Continuously collecting and processing data from network elements to feed into the model.

[0299] Predictive Analysis: Using the trained model to predict potential coverage gaps, congestion hotspots, and other network issues.

[0300] Decision Making: Making real-time decisions on parameter adjustments, such as antenna tilt, power levels, and handover thresholds, to optimize network performance.

[0301] Feedback Loop: Implementing a feedback mechanism to update the model based on the effectiveness of the optimization actions taken.

[0302] The disclosure focuses on enhancing performance through AI / ML models, including potential use cases such as channel state information (CSI) feedback enhancement, beam management, and positioning accuracy improvements.

[0303] The disclosure focus outlines the management capabilities required for AI / ML across different phases: training, validation, testing, deployment, and inference. This includes aspects such as ML entity loading, inference control, and performance evaluation.

[0304] The disclosure utilizes a diverse set of SSB beam shapes with varying beam width configurations and incorporates SSB power boosting. These parameters are dynamically selected and adjusted by an AI / ML optimization algorithm for each sector. The AI / ML algorithm will dynamically choose the appropriate beam shape and optimal SSB power level for each individual SSB beam within the beam-sweeping set per sector, considering both azimuth and elevation angles. This selection process aims to optimize coverage and Signal-to-Interference-plus-Noise Ratio (SINR) while meeting network performance constraints.

[0305] Specifically, the algorithm will assess real-time network conditions to determine the most suitable beamwidth and steering angles. By adjusting the azimuth and elevation angles, the beams can be precisely directed to target-specific user clusters or areas with high traffic demand. This ensures that coverage is maximized, and signal quality is maintained, even in challenging environments such as urban canyons or high-density event venues.

[0306] Additionally, the algorithm will control the power boosting for each SSB beam, fine-tuning the power levels to balance between achieving sufficient coverage and minimizing interference. This dynamic adjustment helps to ensure efficient energy usage and enhances overall network performance. The proposed approach enables a flexible and adaptive network configuration, capable of responding to varying user densities and traffic patterns, thereby optimizing both capacity and user experience.

[0307] FIG. 3 illustrates an exemplary flow diagram of a method (300) for EC prediction based on one or more beamforming actions in the network (108), in accordance with an embodiment of the present disclosure.

[0308] In an embodiment, an EC prediction may be combined with CCO actions. One of the primary aspects of CCO is the dynamic adjustment of transmission power to meet coverage and capacity needs without wasting energy. In a static system, gNBs might operate at a fixed power level, potentially over-providing coverage and capacity, especially during periods of low traffic.

[0309] By integrating CCO, the transmission power of a base station (e.g., gNB) may be adjusted in real-time based on actual demand and network conditions. This means during off-peak hours or in less populated areas, the transmission power can be reduced, thus conserving energy. Conversely, power can be increased only, when necessary, such as during high traffic loads or to support edge users. This targeted use of energy prevents wastage and optimizes power consumption.

[0310] In an embodiment, the EC prediction may be combined with CCO actions including dynamic adjustments on transmitted power. CCO strategies often include the implementation ofsleep modes or low-power states for gNBs during periods of low activity. This involves selectively powering down certain components of the gNB, such as antennas or transmitters, while maintaining minimal functionality for essential operations.

[0311] The ability to switch gNBs or their components into low-power modes during low traffic periods significantly reduces energy consumption. The integration of CCO in the EC prediction function allows the network to dynamically determine when to activate these modes based on real-time data, such as traffic volume and user distribution. The overall reduction in active operational time for gNBs directly translates to lower energy usage.

[0312] In an embodiment, the EC prediction may be combined with CCO actions, including Low-Power mode. Beam management, including dynamic SSB / Traffic beam shaping or beam steering, is another critical technology in 5G that allows gNBs to direct their signals specifically towards user equipment (UE) rather than broadcasting them indiscriminately. CCO includes optimizing beamforming parameters to ensure that coverage and capacity are provided where needed, without excess.

[0313] Optimized beamforming reduces the power needed to cover a given area, especially when fewer UEs are present or in scenarios where precise targeting can minimize interference and energy loss. By managing UE connections more efficiently (e.g., connecting them to the optimal gNB or beam), the network can reduce the power required to maintain quality of service. This precise control and targeting reduce the overall energy consumption of the gNBs.

[0314] In an embodiment, the EC prediction may be combined with CCO actions, including beamforming parameter optimization actions.

[0315] CCO includes advanced traffic prediction capabilities, leveraging AI / ML algorithms to forecast user demand and network conditions. This foresight allows the network to prepare and allocate resources efficiently.

[0316] By predicting traffic patterns, the network can proactively adjust gNB settings in anticipation of changes in demand. For example, during predicted low-traffic periods, gNBs can be preemptively transitioned into low-power states, and transmission power can be lowered. Conversely, during expected peak times, resources can be efficiently ramped up without overshooting, thus preventing unnecessary energy expenditure.

[0317] In an embodiment, the EC prediction may be combined with CCO actions including dynamic traffic predictions.

[0318] CCO considerations are integrated into an EC prediction function for a single gNB in 5G RAN. By leveraging AI / ML-based predictions, dynamic adjustments, and continuous optimization, the algorithm aims to minimize energy consumption while maintaining or improving network performance. The adaptive nature of the algorithm ensures that it can respond effectivelyto real-time network conditions and evolving traffic demands, thereby achieving enhanced energy savings.

[0319] At step (302), the processing unit (120) may be initialized. During this initialization process, a set of input parameters and output parameters may be defined. For example, the input parameters may include maximum transmission power, minimum transmission power, power consumption in sleep mode, a traffic CCO load activation threshold, a traffic offload (mobility) activation threshold, a minimum SSB beam shaping angle width, maximum capacity threshold, and historical for training AI / ML data. In another example, the output parameters may include a baseline energy cost and an optimized energy cost.

[0320] In an embodiment, the processing unit (120) is configured to start SSB beam sweeping on the cell level with the default SSB beamwidth shape. The processing unit (120) may perform SSB beam sweeping of all SSB sweeping beams in a defined time cycle (e.g., 5ms cycle or in configurable time cycle).

[0321] The processing unit (120) is configured to calculate the DL SINR per SSB beam sweeping in conjunction with a cost function. The calculation of DL SINR per SSB sweeping beam with the cost function based on following parameters:- Number of UL PRACH counter attempts of idle UEs- UL CQI reports on connected UEs- HARQ KPI retransmissions (BLER>10%) and- Ll-SINR and Ll-RSRP reporting

[0322] In an aspect, while calculating the cost function, weight coefficients (i.e., weight of each parameter) are considered for all the parameters. The processing engine (130) is configured to build AI / ML algorithm to calculate the cost function for all the parameters. In an aspect, the cost function is calculated to do corrections or fine tuning on the cell shape. Further, the number of UL PRACH counter attempts of idle UEs is calculated from a graph of PRACH attempts versus SINR.

[0323] The processing unit (120) is configured to calculate an SSB beam coverage footprint as a percentage of the default SSB beam shape.

[0324] The processing unit (120) is configured to select the next SSB beamwidth shape until all SSB beamwidth shapes are exhaustively tested and completed. The selection of the SSB beamwidth shape may be configured by an operator.

[0325] The processing unit (120) is configured to select the SSB beam width shape having maximum DL SINR and the coverage footprint > a configurable threshold. The configurable threshold is defined by the operator based on types of services provided by the operator or service requirements from the user (e.g., based on target SINR or target interference to manage thenetwork). The SSB beam width is selected based on the joint contribution of the SINR and the coverage footprint against the threshold.

[0326] For the selected SSB beam width shape, the processing unit (120) is configured to adjust the SSB power-boosting parameters to improve SINR. The DL SINR is calculated based on the selected beamwidth shape. The DL SINR is calculated to check the possibility of improving the SINR. In an aspect, after selecting the SSB beam width shape with maximum DL SINR and the coverage footprint > the configurable threshold, the base station adjusts the SSB power. The base station may receive SINR input from the user equipment. The base station may estimate the SSB beamwidth shape and adjust the SSB power. In this way, the base station may adjust the SSB power to an incremental level.

[0327] The processing unit (120) is configured to perform switching in the coverage area of SSB beams. The processing unit (120) may be configured to inform about changing the beam shape and boosting the SSB beam power to a neighboring cell or nearby base station over an Xn interface. In an aspect, the Xn interface is an interface specified for communication between RAN nodes in standalone operation, such as between the base station (e.g., eNBs and gNBs).

[0328] At step (304), the method (300) may involve performing real-time data collection, which includes the systematic gathering of Real-Time Metrics. In this context, the real-time metrics may encompass various network performance indicators, such as the Synchronization Signal Signal-to-Interference-plus-Noise Ratio (SS-SINR), the Synchronization Signal Reference Signal Received Power (SS-RSRP), and the Synchronization Signal Reference Signal Received Quality (SS-RSRQ). Additionally, the real-time metrics may include various traffic metrics having details of the current traffic load, the number of Radio Resource Control (RRC) connected user equipment (UE), and the distribution of UEs across the network. Furthermore, the real-time metrics may include channel prediction metrics having channel modeling predictions that account for physical obstructions, allowing for enhanced accuracy in forecasting channel conditions and improving overall network performance.

[0329] At step (306), the method (300) may include calculating a baseline energy cost. In an example, the baseline energy cost refers to the standard or reference level of energy consumption associated with the operation of a system or network under typical conditions. In an example, the baseline energy cost may be calculated using the following Equation:

[0330] The baseline EC is calculated based on current power consumption without CCO optimizations, where PCurrent(t) is the power consumption at time t during the time interval of measurements and Δt.

[0331] At step (308), the method (300) may involve AI / ML-based prediction, which encompasses two key areas: traffic and load forecasting and coverage demand analysis. In animplementation, the processing unit (120) may use Al and / or ML algorithms to predict future traffic patterns and load distribution based on historical and real-time data. The processing unit (120) may also predict coverage demand and user mobility patterns.

[0332] At step (310), the method (300) includes dynamic adjustment of network parameters. In examples, the dynamic adjustment of the network parameters includes adaptive transmission power control. In an implementation, the processing unit (120) may perform adaptive transmission power control to calculate optimal transmission power. In an implementation, the processing unit (120) may calculate the optimal transmission power based on predicted traffic load and coverage requirements. In examples, the processing unit (120) may calculate the optimal transmission power using the Equation provided below.Where, Poptrepresents optimal transmission power, Pminrepresents minimum transmission power, Pmaxrepresents maximum transmission power, and Lpred(t) represents the predicted traffic load at time t.

[0333] According to an implementation, the processing unit (120) may perform beamforming optimization. In an implementation, the processing unit (120) may be configured to adjust beamforming angles and widths based on user density and demand. The processing unit (120) may calculate an optimized beam using Equation provided below.Where, Bopt(t)~ may represent optimized beam, Bminrepresents minimum SSB beam shaping angle width, and UEpred(t)' represents the predicted UE distribution.

[0334] In an implementation, the processing unit (120) may activate sleep mode for gNB. In examples, the processing unit (120) may activate the sleep mode based on criteria for sleep mode. In examples, if current load (Lcurrent) is less than the load threshold (Lth), i.e., Lcurrent< Lth, the processing unit (120) may transition the gNB to sleep mode. In example, the sleep mode for the gNB is represented using Equation provided below.

[0335] At step (312), the method (300) includes performing real-time energy cost calculation and optimization. In examples, the processing unit (120) may calculate the optimized energy cost using Equation provided below.where, ECoptrepresents optimized energy cost (EC). The optimized EC is calculated based on the predicted load while ensuring minimum and maximum power constraints.

[0336] In an aspect, the processing unit (120) may be further configured to compare the calculated Optimized EC with the baseline EC. Based on the comparison, the present disclosure may adjust optimization parameters if ECoptis not sufficiently lower than ECbase.

[0337] At step (314), the method (300) includes employing a learning and feedback loop. In an aspect, the processing unit (120) may opt for a performance evaluation, a model refinement, and an optimization refinement for continuous learning and a feedback loop. In the performance evaluation, the present disclosure may evaluate the performance of optimizations in terms of energy savings and network quality. In the model refinement, the present disclosure may update AI / ML models with new data for improved prediction accuracy. In the optimization refinement, the present disclosure may refine CCO strategies and parameters based on performance evaluations and feedback.

[0338] In an embodiment, the EC prediction is employed for uplink CCO, downlink CCO, integrating dynamic beamforming for CCO, and integrating dynamic high-resolution beamforming for CCO, as explained in detail below.

[0339] In an embodiment, an energy cost (EC) prediction algorithm for uplink CCO includes CCO uplink parameters that are implemented with AI / ML-assisted optimization actions. The EC prediction algorithm integrates CCO considerations into an energy cost (EC) prediction function for a single base station (e.g., gNB, gNodeB) in the network to enhance energy savings.

[0340] The EC prediction algorithm for uplink CCO comprises steps, but are not limited to, an initialization step, a real-time monitoring step, a prediction step, a dynamic adjustment step, and a continuous optimization step.

[0341] At the initialization step, input parameters are provided by the network operator. The input parameters comprises, but are not limited to, maximum transmission power (Pmax,UL), minimum transmission power (Pmin,uL)), power consumption in sleep mode (Psleep) traffic load threshold for activating sleep mode (Lth), minimum SSB beam shaping angle widthmaximum capacity threshold (Cmax) and historical data for training AI / ML models. The output parameters baseline energy cost (ECbase) and optimized energy cost (ECopt).

[0342] In an aspect, the Lth is a configurable load threshold parameter. The Lth is configured by the network operator.

[0343] At the real-time data collection step, the real-time metrics are collected from the network. The real-time metrics comprise, but are not limited to, network performance SSB based metrics, traffic metrics and channel prediction. The network performance SSB based metrics comprises signal strength (e.g., SS-SINR, SS-RSRP, SS-RSRQ). The traffic metrics comprisecurrent traffic load (e.g., radio resource control (RRC) connected UEs, UE distribution based on KPIs, counters and real-time statistics from the base station (i.e., gNB Central Unit - User Plane (CU-UP) and Central Unit - Control Plane (CU-CP) in split architecture) OR gNB baseband unit (BBU) in non-split architecture). The channel prediction comprises channel modeling predictions based on physical obstructions, to be used for beam shaping, beam-steering, beam-control and beam-management actions.

[0344] The channel prediction, using models of physical obstructions (e.g., buildings, terrain, mobility), supports multiple key beam-related algorithms. Each beam-related algorithm performs beam actions using the channel prediction.

[0345] At the baseline energy cost calculation step, the baseline EC is calculated based on current power consumption (i.e., at time t, where the time t = current time) without CCO optimizations. The baseline EC is given in Equation below:

[0346] The network operator configures a measurement process window of duration T units, and n discrete At time intervals to take power consumption samples. Hence, the measurement process window = n-At. Therefore, PCurrent(t) is the power consumption at time t during the time interval of measurements At.

[0347] At the AI / ML-based prediction, one or more analysis are performed to predict parameters corresponding to traffic, load, and coverage demand. The one or more analysis comprise a traffic and load forecasting analysis and a coverage demand analysis.

[0348] In the traffic and load forecasting analysis, ML models are used to predict future traffic patterns and load distribution based on historical data (i.e., data before the time t) and realtime data (i.e., data at current measurement period T).

[0349] In the coverage demand analysis, the coverage demand, and user mobility distribution patterns are predicted. The AI / ML algorithm estimates historical statistical data and current UE geo-location to predict the RRC-connected UEs distribution probability in azimuth and elevation angles.

[0350] At the dynamic adjustment of network parameters step, one or more operations are performed. The one or more operations comprise, but are not limited to, adaptive transmission power control, beamforming optimization and sleep mode activation.

[0351] In the adaptive transmission power control, an optimal transmission power is calculated as a min-max optimization function. The optimal transmission power is given in Equation:

[0352] The optimal transmission power is calculated based on predicted traffic load and coverage requirements. The Lpred(t) is the predicted traffic load at the time t. The Pmin,uLfor the uplink UL is the minimum available UL transmitted power by the UE to meet the SINR and pathloss conditions. The Pmax,uL is the constraint of maximum UE power either by cell configuration parameter or UE chipset hardware constraints.

[0353] In the beamforming optimization, beamforming angles and widths are adjusted based on the user density and demand. The beamforming optimization is given by Equation below:Where, UEpred(t) is the predicted UE distribution.

[0354] In the sleep mode activation, if Lcurrent< Lth, then activate transition of the base station (e.g., gNB) to the sleep mode. In the sleep mode, PCurrent(t) = Psleep.

[0355] At the real-time EC calculation and optimization step, the optimized EC is calculated as given in the Equation below.

[0356] In the comparison and adjustment step, the baseline EC ( Cbase) is compared with the optimized EC (ECopt).

[0357] If the ECoptis not sufficiently lower than the ECbase, then the optimization parameters are adjusted. The transmission power and the beamforming parameters (i.e., beamshaping, beam-steering) are adjusted to fulfill the condition ECopt«ECbaseat time period T.

[0358] At the continuous learning and feedback loop step, various steps such as performance evaluation, model refinement, and optimization refinement are performed. In the performance evaluation, the network performance of optimizations is evaluated in terms of energy savings and network quality. In the model refinement, the AI / ML models are updated with new data for improved prediction accuracy. In the optimization refinement, CCO strategies and parameters are refined based on performance evaluations and feedback.

[0359] In another embodiment, an energy cost (EC) prediction algorithm for downlink CCO includes CCO downlink parameters implemented with AI / ML-assisted optimization actions. The EC prediction algorithm integrates CCO considerations into an energy cost (EC) prediction function for a single base station (e.g., gNB, gNodeB) in the network to enhance energy savings.

[0360] The EC prediction algorithm for uplink CCO comprises steps, but are not limited to, an initialization step, a real-time monitoring step, a prediction step, a dynamic adjustment step, and a continuous optimization step.

[0361] At the initialization step, input parameters are provided by the network operator. The input parameters comprise, but are not limited to, maximum transmission power (Pmax,DL), minimum transmission power (Pmin,DL), power consumption in sleep mode (Psleeptraffic load threshold for activating sleep mode (Lth), minimum SSB beam shaping angle widthmaximum capacity threshold (Cmax) and historical data for training AI / ML models.

[0362] In an aspect, the Lth is a configurable load threshold parameter. The Lth is configured by the network operator.

[0363] At the real-time data collection step, the real-time metrics are collected from the network. The real-time metrics comprise, but are not limited to, network performance SSB based metrics, traffic metrics and channel prediction. The network performance SSB based metrics comprises signal strength (e.g., SS-SINR, SS-RSRP, SS-RSRQ). The traffic metrics comprise current traffic load (e.g., radio resource control (RRC) connected UEs, UE distribution based on KPIs, counters and real-time statistics from the base station (i.e., gNB Central Unit - User Plane (CU-UP) and Central Unit - Control Plane (CU-CP) in split architecture) OR gNB baseband unit (BBU) in non-split architecture). The channel prediction comprises channel modeling predictions based on physical obstructions, to be used for beam shaping, beam-steering, beam-control and beam-management actions.

[0364] The channel prediction, using models of physical obstructions (e.g., buildings, terrain, mobility), supports multiple key beam-related algorithms. Each beam-related algorithm performs beam actions using the channel prediction.

[0365] At the baseline energy cost calculation step, the baseline EC is calculated based on current power consumption (i.e., at time t, where the time t = current time) without CCO optimizations. The baseline EC is given in the Equation as below:

[0366] The network operator configures a measurement process window of duration T units, and n discrete At time intervals to take power consumption samples. Hence, the measurement process window = n-At. Therefore, PCurrent(t) is the power consumption at time during the time interval of measurements At.

[0367] At the AI / ML-based prediction step, one or more analysis is performed to predict parameters corresponding to traffic, load, and coverage demand. The one or more analysis comprises a traffic and load forecasting analysis and a coverage demand analysis.

[0368] In the traffic and load forecasting analysis, ML models are used to predict upcoming traffic patterns and load distribution based on historical data (i.e., data before the time t) and real-time data (i.e., data at current measurement period T).

[0369] In the coverage demand analysis, the coverage demand, and user mobility distribution patterns are predicted. The AI / ML algorithm estimates historical statistical data and current UE geo-location to predict the RRC-connected UEs distribution probability in azimuth and elevation angles.

[0370] At the dynamic adjustment of network parameters step, one or more operations are performed. The one or more operations comprise, but are not limited to, adaptive transmission power control, beamforming optimization and sleep mode activation.

[0371] In the adaptive transmission power control, an optimal transmission power is calculated as a min-max optimization function. The optimal transmission power is given in Equation below:

[0372] The optimal transmission power is calculated based on predicted traffic load and coverage requirements. The Lpred(t) is the predicted traffic load at the time t. The Pmin,DLfor the downlink DL is the minimum available DL transmitted power by the base station (e.g., gNB RU unit). The Pmtn,DLis obtained by performing selection of the proper and necessary sub-arrays and RF chains on the mMIMO antenna array to meet the DL SINR and pathloss conditions of the scheduled users. The Pmax,DLis the constraint of maximum DL power of the base station (e.g., gNB RU) given the mMIMO RF chains aggregated power of the antenna panel.

[0373] In the beamforming optimization, beamforming angles and widths are adjusted based on the user density and demand. The beamforming optimization is given by Equation below.Where, UEpred(t) is the predicted UE distribution.

[0374] In the sleep mode activation, if Lcurrent< Lth, then activate transition of the base station (e.g., gNB) to the sleep mode. In the sleep mode, PCurrent(t) = Psleep-

[0375] At the real-time EC calculation and optimization step, the optimized EC is calculated as given in Equation below.

[0376] In the comparison and adjustment step, the baseline EC ECbase) is compared with the optimized EC (ECopt).

[0377] If the ECoptis not sufficiently lower than the ECbase, then the optimization parameters are adjusted. The transmission power and the beamforming parameters (i.e., beamshaping, beam-steering) are adjusted to fulfill the condition ECopt«ECbaseat time period T.

[0378] At the continuous learning and feedback loop step, various steps such as performance evaluation, model refinement, and optimization refinement are performed. In the performance evaluation, the network performance of optimizations is evaluated in terms of energy savings and network quality. In the model refinement, the AI / ML models are updated with new data for improved prediction accuracy. In the optimization refinement, CCO strategies and parameters are refined based on performance evaluations and feedback.

[0379] In another embodiment, an AI / ML-based EC function integrating dynamic beamforming for the CCO is implemented to perform AI / ML-assisted dynamic beam shaping and steering for energy efficient CCO in the network. The AI / ML-based EC function includes realtime dynamic SSB / traffic beam shaping and beam steering as an energy saving mechanism. Further, beam-specific energy profiles control sleep mode activation and the base station (e.g., gNB) power reduction without impacting network service availability. In the AI / ML-based EC function approach emphasizes beam-level fine-tuning as a standalone factor in EC optimization, rather than just focusing on power scaling.

[0380] The AI / ML-based EC function integrating dynamic beamforming for the CCO comprises steps, but are not limited to, an initialization step, a real-time monitoring step, a prediction step, a dynamic adjustment step, and a continuous optimization step.

[0381] At the initialization step, input parameters are provided by the network operator. The input parameters comprise, but are not limited to, maximum transmission power Pmax), minimum transmission powerpower consumption in sleep mode ( Psleep), traffic load threshold for activating sleep mode (Lth), minimum SSB beam shaping angle width (Bmin), maximum capacity threshold (Cmax) and historical data for training AI / ML models. The output parameters are baseline energy cost (ECbase) and optimized energy cost (ECopt).

[0382] In an aspect, the Lthis a configurable load threshold parameter. The Lthis configured by the network operator.

[0383] At the real-time data collection step, the real-time metrics are collected from the network. The real-time metrics comprise, but are not limited to, network performance SSB based metrics, traffic metrics and channel prediction. The network performance SSB based metrics comprises signal strength (e.g., SS-SINR, SS-RSRP, SS-RSRQ). The traffic metrics comprise current traffic load (e.g., radio resource control (RRC) connected UEs, UE distribution based onKPIs, counters and real-time statistics from the base station (i.e., gNB Central Unit - User Plane (CU-UP) and Central Unit - Control Plane (CU-CP) in split architecture) OR gNB baseband unit (BBU) in non-split architecture). The channel prediction comprises channel modeling predictions based on physical obstructions, to be used for beam shaping, beam-steering, beam-control and beam-management actions.

[0384] The channel prediction, using models of physical obstructions (e.g., buildings, terrain, mobility), supports multiple key beam-related algorithms. Each beam-related algorithm performs beam actions using the channel prediction.

[0385] At the baseline energy cost calculation step, the baseline EC is calculated based on current power consumption (i.e., at time t, where the time t = current time) without CCO optimizations. The baseline EC is given in Equation below:

[0386] The network operator configures a measurement process window of duration T units, and n discrete At time intervals to take power consumption samples. Hence, the measurement process window = n-At. Therefore, PCurrent(t) is the power consumption at time t during the time interval of measurements At.

[0387] At the AI / ML-based prediction step, one or more analysis is performed to predict parameters corresponding to traffic, load, and coverage demand. The one or more analysis comprises a traffic and load forecasting analysis and a coverage demand analysis.

[0388] In the traffic and load forecasting analysis, ML models are used to predict upcoming traffic patterns and load distribution based on historical data (i.e. data before the time t) and real-time data (i.e. data at current measurement period T).

[0389] In the coverage demand analysis, the coverage demand, and user mobility distribution patterns are predicted. The AI / ML algorithm estimates historical statistical data and current UE geo-location to predict the RRC-connected UEs distribution probability in azimuth and elevation angles.

[0390] At the dynamic adjustment of network parameters step, one or more operations are performed. The one or more operations comprise, but are not limited to, adaptive transmission power control, beamforming optimization and sleep mode activation.

[0391] In the adaptive transmission power control, an optimal transmission power is calculated as a min-max optimization function. The optimal transmission power is given in Equation below:

[0392] The optimal transmission power is calculated based on predicted traffic load and coverage requirements. The Lpred(t) is the predicted traffic load at the time t.

[0393] In the beamforming optimization, beamforming angles and widths are adjusted based on the user density and demand. The beamforming optimization is given by Equation below:Where, UEpred(t) is the predicted UE distribution and Bmin is the minimum beamwidth allowed by the antenna panel hardware. In an aspect, the predicted UE distribution (UEpred(t)) refers to the forecasted spatial and / or temporal location patterns of the UEs within the network. The predicted UE distribution (UEpred(t)) is generated using historical data, mobility patterns, traffic statistics, or real-time analytics to estimate where and when users are likely to be in the network coverage area.

[0394] An analytical and algorithmic framework performs the beamforming optimization based on beamforming gain, beamwidth, radio frequency (RF) chain / sub -array activation and energy consumption in massive MIMO (mMIMO).

[0395] The analytical and algorithmic framework employs an AI / ML model. The AI / ML model uses the plurality of parameters corresponding to the mMIMO base station. The mMIMO base station comprises one or more antennas divided into one or more sub-arrays. Each sub-array comprises a RF chain, a power amplifier, and a digital baseband path. The plurality of parameters comprises M number of antennas, K number of sub-arrays and total transmit power budget (Pmax).

[0396] A beamforming gain (G) depends on the number of active antenna sub-arrays Nactiveas given in the Equation below:Where, Nactive= Number of active antenna sub-arrays and GBF = Beamforming gain

[0397] The beamforming gain increases with the number of active antenna sub-arrays. As the number of active antenna sub-arrays increases, the beam becomes narrower and concentrates more energy in a specific direction, which enhances the beamforming gain. This results in higher signal power received by the receiving device.

[0398] A beamwidth (0) is inversely proportional to the aperture as given in Equation below:Where, 0 = Beamwidth, X = wavelength, and Daperture = Effective aperture size

[0399] In an aspect, the beamwidth is the angular width of the main lobe of an antenna’s radiation pattern. Aperture refers to the effective size of the antenna array and is proportional to the number of antenna elements and their spacing. The beam becomes narrower as the aperture increases due to more antennas or a larger physical array. Conversely, a smaller aperture results in a wider beam. A larger aperture enables better directional performance by producing narrower beams and higher gain. This allows the antenna to resolve signals from more precise directions, improving spatial selectivity. Narrow beamwidth (i.e., small 0) is desirable in beamforming, as it focuses energy more effectively and reduces interference from unwanted directions.

[0400] An energy cost (EC) function is determined based on one or more parameters corresponding to the antenna sub-arrays. The one or more parameters comprise number of active sub-arrays (Nactive), transmit power per sub-array (Ptx), baseband processing power per active RF chain (Pbb), power amplifier consumption (calculated as function of Ptx), ( ηpa), and minimum required beamforming gain (Greq).

[0401] So, total energy consumption is calculated as given in Equation below:Where, ηpais the amplifier efficiency (e.g., 1 / 0.4)

[0402] One or more constraints need to be considered for the AI / ML-based EC function integrating dynamic beamforming for the CCO. The one or more constraints includes constraint 1 - beamforming gain GBF(Nactive) > Greq, where GBF(Nactive) is beamforming gain of number of active sub-arrays and Greq is minimum required beamforming gain.Nactive-Ptxis total transmit power for the number of active sub-arrays, and P max is maximum power. is total beamwidth for the number of active sub-arrays, and θmax is maximum beamwidth.

[0403] A normalized cost function is calculated as given in Equation below.Where, α,β,γ are tunable weights, E( Nactive, Ptx) is actual energy consumption, (Greq—GBF( Nactive)) is a penalty if gain is below target, and (θ ( Nactive) ~ θtarget) is a penalty for deviating from desired beamwidth.

[0404] An optimized loop algorithm may be applied to obtain an optimal energy-optimized beam configuration in the mMIMO. The optimized loop algorithm comprises following steps:

[0405] Input parameters comprise the total transmit power (Pmax), the minimum required beamforming gain (Greq), the desired beamwidth θtarget and total subarrays (M). The Nactiveis 1 to M. For each Nactive, G_BF (Nactive) and θ(Nactive) are calculated. The GBF(Nactive) is calculated as 10 log10(Nactive). The θ(Nactive) is calculated as Ptxis calculated as min{Pmax / Nactive, required per-user power} .

[0406] For each (Nactive, Ptx), the Etotai and the EC are calculated. The Etotai is calculated as given in Equation below:Etotal=Nactive* (Pbb+ ηpa* Ptx)

[0407] The cost C(Nactive, Ptx) is evaluated for each (Nactive, Ptx).

[0408] (Nactive, Ptx) = argmin C is selected from calculated cost C for each (Nactive, Ptx). The selected (Nactive, Ptx) is an optimal energy-optimized beam configuration i.e., optimal (Nactive, Ptx).

[0409] In the sleep mode activation, If LCurrent< Lth, then activate transition of the base station (e.g., gNB) to the sleep mode. In the sleep mode, Pcurrent(t) = Psieep.

[0410] In an aspect, an expanded energy-aware beamforming with the sleep mode is performed based on an instantaneous cell load (0 to 1) (Lcurrent), load threshold below which the system enters sleep mode (Lth), minimal number of sub-arrays to form a low-power wide beam ( Nsleep) and reduced gain and increased beamwidth during sleep mode (Gsieep).

[0411] If the cell load is low, the base station is switched to a low-power configuration. The Nsieep sub-arrays are activated in the low-power configuration. Reduced beamforming gain (Gsieep) is accepted, and broad coverage (wide beamwidth) is maintained. Full precision beams are resumed when the load exceeds the load threshold.

[0412] In load-aware energy optimization with sleep mode in the mMIMO, the input parameters comprise Pmax, Greq, θtarget, M total sub-arrays, LCurrent, and Lth.

[0413] The Lcurrent (real-time cell load) is monitored to detect whether LCurrent< Lth. Upon detecting Lcurrent < Lth, the sleep mode is activated. The activation of the sleep mode comprises the following steps:

[0414] Settings are performed such as Nactive= Nsleep(e.g., down to 4 sub-arrays), beam to wide beam with θsleep, GBF = 10 1ogl 0(Nsleep) and the reduced Ptx= min(Pmax / Nsleep, required power).

[0415] Etotal_sleep is calculated as given in Equation below:Etotal_sleep=Nsleep* (Pbb+ ηpa* Ptx)

[0416] Low-energy wide-beam is applied.

[0417] The condition LCurrent> Lthis periodically checked to resume normal mode.

[0418] Upon detecting LCUrrent > Lth, Nactive= 1 to M is initialized.

[0419] For each Nactive, GBF and 0 are computed as given in the Equations below:GBF - l O logl O(Nactive)

[0420] Minimum Ptxis determined based on required user QoS

[0421] Energy (E) is calculated as given in the Equation below:E — Nactive* (Pbb+ ηpa* Ptx)

[0422] The cost (C) is computed as given in the Equation below:

[0423] (Nactive*, Ptx*) = argmin C is selected. The expression (Nactive*, Ptx*) = argmin C represents determining of optimal number of active antenna sub-arrays and the optimal transmission power that minimizes the cost function. So, the selecting number of active antenna sub-arrays and the transmit power that minimizes the cost function. Based on the selection, the optimal beam configuration is applied.

[0424] At the real-time EC calculation and optimization step, the optimized EC is calculated as given in the Equation below.

[0425] In the comparison and adjustment step, the baseline EC (ECfease) is compared with the optimized EC (ECopt).

[0426] If the ECoptis not sufficiently lower than the ECbase, then the optimization parameters are adjusted. The transmission power and the beamforming parameters (i.e., beamshaping, beam-steering) are adjusted to fulfill the condition ECopt«ECbaseat time period T.

[0427] At the continuous learning and feedback loop step, various steps such as performance evaluation, model refinement, and optimization refinement are performed. In the performance evaluation, the network performance of optimizations is evaluated in terms of energysavings and network quality. In the model refinement, the AI / ML models are updated with new data for improved prediction accuracy. In the optimization refinement, CCO strategies and parameters are refined based on performance evaluations and feedback.

[0428] In another embodiment, an AI / ML-based EC function integrating dynamic high- resolution beamforming for the CCO is implemented to perform AI / ML-assisted dynamic beam shaping and steering for energy efficient CCO in the network. The AI / ML-based EC function includes real-time dynamic SSB / traffic beam shaping and beam steering as an energy saving mechanism. The high-resolution beam prediction models determine beam directivity and width based on UE clustering and interference conditions. In an aspect, the high-resolution beam refers to a narrow, precisely directed beam formed using advanced beamforming techniques, in massive Multiple Input, Multiple Output (mMIMO) systems. The high-resolution beams have narrow beamwidth, high spatial precision to allow fine control over where the radio energy is sent, and high directivity to focus energy on a specific user or small area. The high-resolution improves capacity to enable spatial multiplexing by allowing multiple narrow beams to serve different users at the same time-frequency resource, enhances coverage to reach users in hard-to-cover areas by precisely steering beams and reduces interference to minimize signal leakage into unintended directions.

[0429] Further, beam-specific energy profiles control sleep mode activation and the base station (e.g., gNB) power reduction without impacting network service availability. In the AI / ML- based EC function approach emphasizes beam-level fine-tuning as a standalone factor in EC optimization, rather than just focusing on power scaling.

[0430] The AI / ML-based EC function integrating dynamic high-resolution beamforming for the CCO comprises steps, but are not limited to, initialization step, a real-time monitoring step, a prediction step, a dynamic adjustment step, and a continuous optimization step.

[0431] At the initialization step, input parameters are provided by the network operator. The input parameters comprise, but are not limited to, maximum transmission power Pmax), minimum transmission power (Pmm), power consumption in sleep mode ( Psleep) traffic load threshold for activating sleep modeminimum SSB beam shaping angle widthmaximum capacity threshold Cmax) and historical data for training AI / ML models. The output parameters are baseline energy cost (ECbase) and optimized energy cost (ECopt).

[0432] In an aspect, the Lthis a configurable load threshold parameter. The Lthis configured by the network operator.

[0433] At the real-time data collection step, the real-time metrics are collected from the network. The real-time metrics comprise, but are not limited to, network performance SSB based metrics, traffic metrics and channel prediction. The network performance SSB based metrics comprises signal strength (e.g., SS-SINR, SS-RSRP, SS-RSRQ). The traffic metrics comprisecurrent traffic load (e.g., radio resource control (RRC) connected UEs, UE distribution based on KPIs, counters and real-time statistics from the base station (i.e., gNB Central Unit - User Plane (CU-UP) and Central Unit - Control Plane (CU-CP) in split architecture) OR gNB baseband unit (BBU) in non-split architecture). The channel prediction comprises channel modeling predictions based on physical obstructions, to be used for beam shaping, beam-steering, beam-control and beam-management actions.

[0434] The channel prediction, using models of physical obstructions (e.g., buildings, terrain, mobility), supports multiple key beam-related algorithms. Each beam-related algorithm performs beam actions using the channel prediction.

[0435] At the baseline energy cost calculation step, the baseline EC is calculated based on current power consumption (i.e., at time t, where the time t = current time) without CCO optimizations. The baseline EC is given in Equation below.

[0436] The network operator configures a measurement process window of duration T units, and n discrete At time intervals to take power consumption samples. Hence, the measurement process window = n-At. Therefore, PCurrent(t) is the power consumption at time t during measurement time interval At.

[0437] At the AI / ML-based prediction step, one or more analysis is performed to predict parameters corresponding to traffic, load, and coverage demand. The one or more analysis comprises a traffic and load forecasting analysis and a coverage demand analysis.

[0438] In the traffic and load forecasting analysis, ML models are used to predict upcoming traffic patterns and load distribution based on historical data (i.e., data before the time t) and real-time data (i.e., data at current measurement period T).

[0439] In the coverage demand analysis, the coverage demand, and user mobility and distribution patterns are predicted. The AI / ML algorithm estimates historical statistical data and current UE geo-location to predict the RRC-connected UEs distribution probability in azimuth and elevation angles.

[0440] At the dynamic adjustment of network parameters step, one or more operations are performed. The one or more operations comprise, but are not limited to, adaptive transmission power control, beamforming optimization and sleep mode activation.

[0441] In the adaptive transmission power control, an optimal transmission power is calculated as a min-max optimization function. The optimal transmission power is given in Equation

[0442] The optimal transmission power is calculated based on predicted traffic load and coverage requirements. The Lpred(t) is the predicted traffic load at the time t.

[0443] In the beamforming optimization, beamforming angles and widths are adjusted based on the user density, angular distribution, throughput, demand and energy consumption with high-resolution beam gain. The beamforming optimization is given by the following Equation:

[0444] Where, UEprecl(t)' is the predicted UE angular and spatial density and g(-) is function mapping UE distribution to beamwidth. In an aspect, the predicted UE angular and spatial density (t / Epred(t)) refers to the forecasted concentration of the UEs in the network, measured in both angular domain (i.e., direction relative to the base station), and spatial domain (i.e., physical location within the coverage area). The predicted UE angular and spatial density is used to estimate how many UEs are expected to be located in specific directions and positions, based on historical data, mobility trends, and real-time network analytics. In an aspect, the g(-) function mapping UE distribution to the beam is used to decide how wide the beam should be, depending on how users are distributed. For example, if the users are clustered, a narrow beam is used, and if users are spread out, a wider beam is used.

[0445] High-resolution (also termed as super-resolution) improves the effective beam gain and angular resolution with fewer antennas. To improve effective beam gain and angular resolution with fewer antennas, the high-resolution beamforming uses a super-resolution gain factor. In an aspect, the super-resolution gain factor refers to a metric that quantifies the improvement in resolution achieved by the super-resolution beamforming compared to conventional beamforming.

[0446] In high-resolution beamforming for the CCO, the super-resolution gain factor (\| / SR) is considered. Let considerbe the super-resolution factor which represents the gain in angular resolution compared to classical beamwidth (Oclassical) as given in the Equation below:

[0447] Narrower beams (i.e., beams narrower than standard array limits) are created with the same number of active antenna sub-arrays. The narrower beam is given in the Equation below.

[0448] An analytical and algorithmic framework performs the beamforming optimization based on beamforming gain, beamwidth, radio frequency (RF) chain / sub -array activation and energy consumption in massive MIMO (mMIMO).

[0449] The analytical and algorithmic framework employs an AI / ML model. The AI / ML model uses the plurality of parameters corresponding to the mMIMO base station. The mMIMO base station comprises one or more antennas divided into one or more sub-arrays. Each sub-array comprises a RF chain, a power amplifier, and a digital baseband path. The plurality of parameters comprises M number of antennas, K number of sub-arrays and total transmit power budget (Pmax).

[0450] A beamforming gain (G) depends on the number of active antenna sub-arrays Nactiveas given in the Equation below:Where, Nactive= Number of active antenna sub-arrays Nactive) and is super resolution gainfactor.

[0451] The beamforming gain increases with the number of active antenna sub-arrays. As the number of active antenna sub-arrays increases, the beam becomes narrower and concentrates more energy in a specific direction, which enhances the beamforming gain. This results in higher signal power received by the receiving device.

[0452] A bandwidth (0) is inversely proportional to the aperture as given in the Equation below:Where, θ = Beamwidth, λ = wavelength, and Daperture = Effective aperture size and Nactivenumber of active antenna sub-arrays.

[0453] Wider beams, achieved by activating a reduced number of antennas, resulting in lower energy consumption but correspondingly lower beamforming gain. Consequently, to maintain the desired communication quality, a higher transmit power is required.

[0454] An energy cost (EC) function is determined based on one or more parameters corresponding to the antenna sub-arrays. The one or more parameters comprise number of active sub-arrays (Nactive), transmit power per sub-array (Ptx), baseband processing power per active RFchain (Pbb), power amplifier consumption (calculated as function of Ptx) (Ppa), and minimum required beamforming gain (Greq).

[0455] So, total energy consumption is calculated as given in the Equation belowWhere, ηpais the amplifier efficiency (e.g., 1 / 0.4)

[0456] One or more conditions need to be considered for the AI / ML-based EC function integrating dynamic beamforming for the CCO. The one or more conditions are as follows:

[0457] Condition 1 - beamforming gain where GBF(Nactive) isbeamforming gain of number of active sub-arrays and Greq is minimum required beamforming gain. GBF(Nactive) is greater than or equal to the required gain Greq. Condition 1 i.e.,ensures that the system supports the current load without dropping below the minimum service quality.

[0458] is total transmit power for the number of active sub-arrays, and Pmax is maximum power. The total transmit power (i.e., transmit power per sub-array*number of active sub-arrays) is less than or equal to the maximum power. Condition 2 i.e.,ensures that total power consumed by the active sub-arrays stays within the system’s maximum power allowed limit.

[0459] is total beamwidth for the number of activesub-arrays, and θmax is maximum beamwidth. The total beamwidth for the number of active subarrays is less than or equal to the maximum beamwidth. Condition 3 - 0(Nactive) < θmax ensures that the beamwidth stays within maximum beamwidth for system performance, security, and hardware capabilities.

[0460] A normalized cost function is calculated as given in the Equation below.Where,are tunable weights, is actual energy consumption, is a penalty if gain is below target, isa penalty for deviating from desired beamwidth and is used to fine tune complex super-resolution settings (which imply higher digital signal processing (DSP) cost or latency). In an aspect, for high resolution beamforming for CCO, super resolution gain factor isconsidered during calculation of the normalized cost function.

[0461] An optimized loop algorithm may be applied to obtain an optimal energy-optimized beam configuration in the mMIMO. The optimized loop algorithm comprises the following steps:

[0462] Input parameters comprise the total transmit power (Pmax), the minimum required beamforming gain (Greq), the desired beamwidth θtarget and total subarrays (M) and the maximum super-resolution factor (\| / sR,max).

[0463] For each Nactive, GBF and 0 are calculated. The Nactiveis 1 to M and the to(e g-, 1 0 to 2.0 in 0.1 steps).

[0464] The GBF is calculated as 10 1ogl 0(Nactive, \| / SR). The 6 is calculated asPtxis calculated as Ptx= min{Pmax / Nactive, required per-user power}.

[0465] For each (Nactive, Ptx), the Etotai and the EC are calculated. The Etotai is calculated as given in Equation below

[0466] The total energy consumption is calculated based on the number of active antenna sub-arrays, the baseband processing power, the transmit power and the high-resolution gain factor.

[0467] The cost is evaluated as given in the Equation below:

[0468] The cost C is calculated by considering the super resolution gain factor

[0469] Expression = argmin C indicates select number of activeantenna sub-array (Nactive), transmit power (Ptx), and super resolution gain factor thatminimizes the cost C. The selected is stored. The selected (N_active, P tx,'SR) is an optimal energy-optimized beam configuration i.e., optimal

[0470] In the sleep mode activation, if then activate transition of the basestation (e.g., gNB) to the sleep mode. In the sleep mode, PCurrent( ) = Psleep-

[0471] In an aspect, an expanded energy-aware beamforming with the sleep mode is performed based on an instantaneous cell load (0 to 1) Lcurrent), load threshold below which the system enters sleep mode (Lth), minimal number of sub-arrays to form a low-power wide beam (Nsleep) and reduced gain and increased beamwidth during sleep mode (Gsleep).

[0472] If the cell load is low, the base station is switched to a low-power configuration. The Nsleepsub-arrays are activated in the low-power configuration. Reduced beamforming gain(Gsieep) is accepted, and broad coverage (wide beamwidth) is maintained. Full precision beams are resumed when the load exceeds the load threshold.

[0473] In load-aware energy optimization with sleep mode in the mMIMO, the input parameters comprise Pmax, Greq, θtarget, M total sub-arrays, Lcurrent, Lthand SR.max-

[0474] The Lcurrent(real-time cell load) is monitored to detect whether Lcurrent< Lth. Upon detecting Lcurrent< Lth, the sleep mode is activated. The activation of the sleep mode comprises following steps:

[0475] Settings are performed such as Nactive= Nsleep(e.g., down to 4 sub-arrays), beam to wide beam with θsleep, GBF = 10 1ogl 0(Nsleep) and the reduced Ptx= min(Pmax / Nsleep, required power).

[0476] Etotai_sleep is calculated as given in the Equation below:Etotal_sleep=Nsleep* (Pbb+ ηpa* Ptx)

[0477] Low-energy wide-beam is applied.

[0478] The condition Lcurrent> Lth is periodically checked to resume normal mode.

[0479] Upon detecting Lcurrent> Lth, Nactive= 1 to M is initialized.

[0480] For each Nactive, GBF and 0 are computed as given in Equations below:GBF = 10 log10(Nactive)

[0481] Minimum Ptxis determined based on required user QoS

[0482] Energy (E) is calculated as given in the Equation belowE = Nactive* (Pbb+ ηpa* Ptx)

[0483] The cost (C) is computed as given in the Equation below:

[0484] is selected. Based on the selection, the optimal beam configuration is applied.

[0485] At the real-time EC calculation and optimization step, the optimized EC is calculated as given in the Equation below:

[0486] In the comparison and adjustment step, the baseline EC (ECbase) is compared with the optimized EC (ECopt).

[0487] If the ECoptis not sufficiently lower than the ECbase, then the optimization parameters are adjusted. The transmission power and the beamforming parameters (i.e., beamshaping, beam-steering) are adjusted to fulfill the condition ECopt«ECbaseat time period T.

[0488] At the continuous learning and feedback loop step, various steps such as performance evaluation, model refinement, and optimization refinement are performed. In the performance evaluation, the network performance of optimizations is evaluated in terms of energy savings and network quality. In the model refinement, the AI / ML models are updated with new data for improved prediction accuracy. In the optimization refinement, CCO strategies and parameters are refined based on performance evaluations and feedback.

[0489] The present disclosure introduces an energy-effi cient strategy for 5G and beyond RAN systems, where Coverage and Capacity Optimization (CCO) actions and traffic offloading are treated as mutually exclusive yet cooperative mechanisms within a unified Energy Cost (EC) prediction and control framework. Advanced CCO techniques (e.g., high-resolution beamforming, SSB / traffic beam-shaping, and beam steering) significantly reduce energy consumption by dynamically targeting high-demand areas with optimized directional coverage. These enhancements may eliminate traffic offloading, enabling the serving network node (e.g., a base station such as a gNB) to deliver energy-efficient service without exceeding power budgets.

[0490] By decoupling CCO actions from offloading procedures, the network gains interpretability and reduces the complexity of energy optimization. In conventional architecture, the joint tuning of beam parameters and load balancing introduces parameter interdependencies and decision-making conflicts. The proposed approach enables a simpler, modular control logic, where CCO actions are prioritized, and offloading is considered only when CCO is insufficient or ineffective, serving as a fallback mechanism.

[0491] Furthermore, EC prediction at each NG-RAN node becomes more accurate and actionable when beamforming improvements are evaluated independently of traffic migration. Since offloading introduces external load variability and dynamic resource shifts between nodes, it inherently complicates EC forecasting. A CCO-first strategy offers a stable and consistent foundation for predicting energy savings particularly in scenarios where localized beam control adapts to UE traffic behavior (e.g., bursty or intermittent patterns), enabling sleep-mode activations or sub-array deactivations.

[0492] The present disclosure also presents an advanced EC-aware coordination mechanism, in which a serving network node (e.g., gNB) intending to offload UEs shares a load profile with a candidate target node. The target node responds with a predicted EC delta based on its ability to absorb the load efficiently via its CCO capabilities. If the combined EC is reduced, the offloading proceeds; otherwise, beam adjustments are applied at the source node. This internode coordination ensures that offloading only occurs when the resulting energy profile across all involved nodes is optimal.

[0493] By treating CCO and offloading as mutually exclusive steps within a single algorithm, the system achieves greater decision clarity, reduced control complexity, and interpretable EC evaluations. The present disclosure supports a distributed, energy-aware RAN control plane, aligned with sustainability objectives and scalable automation in next-generation networks.

[0494] The present disclosure proposes a combination of SSB beam shaping, SSB beam boosting, and energy cost prediction with potential offloading actions, thus enabling more sustainable and efficient 5G network operation. This integrated approach facilitates dynamic adjustments to beamforming strategies, optimizing both coverage and capacity while aligning with energy-saving and processing load goals. By combining CCO with potential neighboring cell offloading (mobility management) into a single Energy Cost (EC) prediction function, this approach significantly enhances energy savings in 5G Radio Access Networks (RAN) for several reasons. This integrated approach allows for more holistic and dynamic management of network resources, leading to efficient energy usage while maintaining the desired service quality. The incorporation of an energy cost prediction function into the CCO framework represents a significant advancement toward sustainable network management. Additionally, this approach ensures the efficient use of energy resources and aligns with global efforts to reduce carbon footprints and improve traffic load distribution. As 5G networks continue to expand and evolve, such proactive energy management strategies with offloading actions are essential in balancing technological advancements with environmental and economic considerations.

[0495] The present disclosure allows for the dynamic adaptation of network resources based on real-time conditions, thereby reducing unnecessary energy usage. Further, the present disclosure enhances the granularity of energy management, enabling more precise control over which network elements are active at any given time. Also, the present disclosure supports the implementation of network strategies, such as turning off redundant cells during low-demand periods or optimizing the energy efficiency of active cells. For example, integrating CCO considerations with the potential offloading of traffic to neighboring cells into the EC prediction function for a single gNodeB (gNB) in a 5G network offers energy-saving enhancements. The integration of CCO involves optimizing the network's coverage and capacity, directly impacting energy consumption. By distributing the traffic load more evenly across gNBs, the network prevents situations where some gNBs are overburdened while others are underutilized. Underutilized gNBs are placed into low-power modes or even turned off, reducing overall energyconsumption. Efficient load balancing ensures that no single gNB is pushed to operate at full capacity unnecessarily, thus conserving energy across the network.

[0496] The combined use of EC prediction and intelligent offloading is essential in balancing high-performance demands with sustainable energy consumption. In conclusion, the amalgamation of the EC prediction function with offloading strategies represents a critical advancement in managing dense 5G and future 6G networks. This approach enables significant energy savings and supports the efficient utilization of network resources, ensuring that the benefits of dense deployments do not come at the expense of excessive energy consumption. By leveraging predictive analytics and adaptive handover mechanisms, this integrated strategy helps achieve a sustainable balance between network performance and energy efficiency, which is crucial for the future of telecommunications infrastructure.

[0497] The present disclosure involves an algorithm that dynamically selects the appropriate beam shape among the available beam width beams and the most suitable SSB power per individual SSB beam of the beam-sweeping set per sector. This selection process operates under the constraints of desired energy savings and potential offloading actions, aiming to sustain or improve coverage and Signal-to-Interference-plus-Noise Ratio (SINR). This method ensures that network performance is maintained or enhanced while achieving energy efficiency goals, making it a pivotal development in modern network management.

[0498] In a communication network, Massive MIMO (Multiple Input Multiple Output) antenna arrays enable a network node, such as a gNodeB (gNB), to enhance beamforming capabilities by dynamically forming a radiation pattern that amplifies radiated power in specific directions. This focused transmission optimizes signal strength at user equipment, thereby reducing interference and improving overall communication quality. By effectively directing energy rather than broadcasting uniformly, the network maximizes the utilization of available spectrum, which is particularly crucial in 5G mid-band Time Division Duplex (TDD) deployments, where efficient spectrum management is necessary to accommodate both uplink and downlink traffic.

[0499] There are, however, situations where the use of broader beamwidth beams is advantageous for various system functions, including synchronization, initial access, and mobility signaling. These functions are particularly relevant for control and broadcast channels, which are designed to disseminate information to multiple users simultaneously. The broader beamwidth beams facilitate a larger coverage area, ensuring that signals reach a wider range of devices, thereby enhancing the reliability of synchronization processes and simplifying initial access for users connecting to the network. Additionally, since control and broadcast channels typically handle information with a broadcast nature and a low bitrate per user, the broader beams allow for effective communication without the need for high precision in directionality, resulting in improved network efficiency and user experience, especially in scenarios with high user density or mobility.1

[0500] For large antenna arrays, creating a broad beam is not straightforward. There are several different ways, like transmitting a small sub-group of antenna dipole elements (sub-arrays) with a broad beamwidth radiation pattern or using a tunable beam-shaping methodology to broaden the beamwidth and preserve the power. An alternative and efficient tunable beam-shaping methodology constitutes a common approach used for beam management procedures in 5G New Radio (NR), relying on performing a beam-sweep over a set of SSB beams to cover the desired cell area in a time-multiplexed manner fully. Hence, instead of constantly transmitting a single SSB broad beam, multiple SSBs are sequentially transmitted in what is known as a “synchronization signal burst” containing a set of narrower beams spanning the sector's angular (and perhaps the elevation) directions. However, this method is prone to a major disadvantage that multi-SSB beam shapes with specific beamwidth configurations are fixed across all the sites based on the antenna equipment capability. This leads to unnecessary transmissions to undesired directions, overlapping sector areas and inter-cell interference across neighbor dense sectors with high energy consumption and lower energy efficiency.

[0501] In an embodiment, for AI / ML based CCO, for the existing SSB beam sweeping functionality on serving node, different SSB beams with specific SSB beam- width configurations along with SSB beam steering azimuth and elevation angles are dynamically selected to optimize coverage and SINR.

[0502] The goal is to minimize the intra-cell and inter-cell beam overlapping, preserve the sector coverage and maximize the expected SINR for different inter-site distance and geography. Hence, ultimately the multi-SSB beam-width beams will improve the DL coverage and throughput without affecting the UE capability.

[0503] In an embodiment, for AI / ML based CCO, the proposed algorithm refers to Pl beam management algorithm, hence the user device is unaffected and follows the specific 3 GPP and GSMA.

[0504] The proposed algorithm, to support AI / ML in NG-RAN, gets inputs only from connected UEs in the serving cell, building a cost function to predict the SSB coverage per individual SSB beam width out of the existing SSB beam sweeping beams.

[0505] In an embodiment, for AI / ML based CCO, the inference input may be collected from UE or local node or neighbour node, e.g. including:

[0506] Inputs from connected UEs as follows:- connected UE measurement report (e.g., LI SS-RSRP, SS-RSRQ, SS-SINR measurement, CQI, etc), including cell level and beam level UE measurements- connected UE performance (HARQ NACK KPI retransmissions (BLER>10%)),- connected UE location information (e.g., coordinates, serving cell ID)- idle mode UE’s Random Access performance (RACH success rate, number of RACH attempts) to convert to SINR from existing graph of number of PRACH attempts vs SINR).- connected UE RLF report

[0507] The future coverage status may be Cell Coverage Modification(s) (i.e., power parameters, SSB boosting etc.) or SSB Coverage Modification(s) (i.e., SSB beam width modifications).

[0508] In an embodiment, for AI / ML based CCO, the future coverage status may include:Cell Coverage Modification including SSB beam power boosting, for the Cell Coverage State which is used to solve the predicted CCO issue of the certain predicted cell;SSB Coverage Modification including SSB beam width and SSB beam steering Coverage State which is used to solve the predicted CCO issue of the certain predicted SSB.

[0509] DL performance is investigated based on SSB power boosting and SSB beam- width shaping for different SSB beam gains on the beam-sweeping procedure, building a proper cost function.

[0510] Goal is to improve (increase) DL SINR and average DL cell throughput

[0511] In an embodiment, for AI / ML based CCO, the future coverage status may include the following NG-RAN gNB or gNB-CU / gNB-DU algorithm steps:

[0512] Step 1 : Initial SSB Beam Sweeping

[0513] Start SSB beam sweeping at the cell level with the default (operator-configurable) SSB beam- width shape. Perform sweeping for all SSB beams within a 5ms cycle.

[0514] Step 2: SINR Estimation and Data Collection

[0515] Estimate the Downlink (DL) SINR per SSB sweeping beam using a combined cost function. This includes:- Number of Uplink (UL) PRACH counter attempts (build a graph of PRACH attempts vs.SINR) for idle UEs.- Connected UEs' UL Channel Quality Indicator (CQI) reports.- HARQ KPI retransmissions (Block Error Rate > 10%).- LI -SINR and LI -Reference Signal Received Power (RSRP) reporting.

[0516] Step 3: Beam Steering and Beam-width Shape Testing

[0517] For each SSB beam-width shape, dynamically adjust the SSB beam's azimuth and elevation angles. Repeat Step 1 and Step 2 for each configuration, ensuring all combinations of beam-width shapes and steering angles are exhaustively tested.

[0518] Step 4: Coverage Footprint Estimation

[0519] For each tested configuration, estimate the SSB beam coverage footprint as a percentage of the default SSB beam shape.

[0520] Step 5: Selection of Optimal Beam Configuration

[0521] Identify the configuration with the argmax DL UL SINR performance that also meets a predefined coverage footprint threshold (configurable). This configuration includes the optimal combination of beam- width shape, azimuth / elevation angles, and SSB power level.

[0522] Step 6: Fine-tuning with SSB Power Boosting

[0523] For the selected optimal beam-width shape and steering angles, adjust the SSB power boosting parameters to further enhance SINR. Re-estimate the DL SINR based on the refined settings.

[0524] Step 7: Final Configuration Application

[0525] Apply the optimal beam configuration (beam-width shape, steering angles, and power boosting) to the network for active use, ensuring maximum coverage and optimal signal quality.

[0526] Step 8: Continuous Monitoring and Adaptation

[0527] Continuously monitor the network's performance metrics, such as SINR, coverage, and user experience indicators. Use AI / ML algorithms to dynamically adjust the configurations in response to changing network conditions and traffic patterns.

[0528] In an embodiment, an energy cost prediction should be combined with CCO actions.

[0529] One of the primary aspects of CCO is the dynamic adjustment of transmission power to meet coverage and capacity needs without wasting energy. In a static system, gNBs might operate at a fixed power level, potentially over-providing coverage and capacity, especially during periods of low traffic.

[0530] By integrating CCO, the transmission power of a gNB can be adjusted in real-time based on actual demand and network conditions. This means during off-peak hours or in less populated areas, the transmission power can be reduced, thus conserving energy. Conversely,power can be increased only, when necessary, such as during high traffic loads or to support edge users. This targeted use of energy prevents wastage and optimizes power consumption.

[0531] In an embodiment, an energy cost prediction should be combined with CCO actions including dynamic adjustments on transmitted power.

[0532] CCO strategies often include implementing sleep modes or low-power states for gNBs during periods of low activity. This involves selectively powering down certain components of the gNB, such as antennas or transmitters, while maintaining minimal functionality for essential operations.

[0533] The ability to switch gNBs or their components into low-power modes during low traffic periods significantly reduces energy consumption. The integration of CCO in the EC prediction function allows the network to dynamically determine when to activate these modes based on real-time data, such as traffic volume and user distribution. The overall reduction in active operational time for gNBs directly translates to lower energy usage.

[0534] In an embodiment, an Energy Cost prediction should be combined with CCO actions, including Low-Power mode.

[0535] Beam management including dynamic SSB / Traffic beam shaping or beam steering is another critical technology in 5G that allows gNBs to direct their signals specifically towards user equipment (UE) rather than broadcasting them indiscriminately. CCO includes optimizing beamforming parameters to ensure that coverage and capacity are provided where needed, without excess.

[0536] Optimized beamforming reduces the power needed to cover a given area, especially when fewer UEs are present or in scenarios where precise targeting can minimize interference and energy loss. By managing UE connections more efficiently (e.g., connecting them to the optimal gNB or beam), the network can reduce the power required for maintaining quality of service. This precise control and targeting reduce the overall energy consumption of the gNBs.

[0537] In an embodiment, an Energy Cost prediction should be combined with CCO actions including beamforming parameter optimization actions.

[0538] CCO includes advanced traffic prediction capabilities, leveraging AI / ML algorithms to forecast user demand and network conditions. This foresight allows the network to prepare and allocate resources efficiently.

[0539] By predicting traffic patterns, the network can proactively adjust gNB settings in anticipation of changes in demand. For example, during predicted low-traffic periods, gNBs can be preemptively transitioned into low-power states, and transmission power can be lowered.Conversely, during expected peak times, resources can be efficiently ramped up without overshooting, thus preventing unnecessary energy expenditure.

[0540] In an embodiment, an Energy Cost prediction should be combined with CCO actions including dynamic traffic predictions.

[0541] An algorithm which outlines a systematic approach to integrating CCO considerations into an EC prediction function for a single gNB in 5GRAN. By leveraging AI / ML- based predictions, dynamic adjustments, and continuous optimization, the algorithm aims to minimize energy consumption while maintaining or improving network performance. The adaptive nature of the algorithm ensures that it can respond effectively to real-time network conditions and evolving traffic demands, thereby achieving enhanced energy savings.

[0542] In an embodiment, an Energy Cost prediction should be combined with CCO actions including potential offloading action to neighbor cells.

[0543] Hence, combining Coverage and Capacity Optimization (CCO) with potential neighbor cell offloading (mobility management) into a single Energy Cost (EC) prediction function can significantly enhance energy savings in 5G Radio Access Networks (RAN) for several reasons. This integrated approach allows for more holistic and dynamic management of network resources, leading to more efficient energy usage while maintaining the desired service quality.

[0544] The following are some key reasons and the corresponding proposals for this combined CCO and offloading functionality:

[0545] Dynamic Load Balancing and Traffic Distribution:

[0546] In a 5G network, traffic loads may vary significantly across different cells and times of the day. Some cells may become congested, while others remain underutilized.

[0547] By incorporating neighbor cell offloading into the EC function, the network can dynamically distribute traffic from congested cells to underutilized neighboring cells. This redistribution helps prevent excessive power usage in heavily loaded cells and allows for potential power reductions or sleep mode activation in less utilized cells, thus optimizing overall energy consumption.

[0548] In an embodiment, an Energy Cost prediction should include an enhanced traffic distribution offloading action to improve energy efficiency.

[0549] In Adaptive Coverage and Capacity Management, Coverage and capacity requirements fluctuate based on user density and demand patterns. For instance, during special events or peak hours, higher capacity and coverage might be needed, whereas off-peak hours mightallow for reductions. Moreover, mobility management, including handovers and load balancing, is crucial in a mobile network to maintain service quality. However, inefficient handling may lead to unnecessary power consumption, especially if users frequently move between cells with differing loads.

[0550] The combined consideration of CCO and mobility in the EC function enables adaptive adjustments in both coverage (by controlling transmission power and antenna tilts) and capacity (by adjusting resource allocation and scheduling). This means that the network may minimize energy usage by reducing unnecessary coverage areas and reallocating resources more efficiently based on real-time demand.

[0551] In an embodiment, an Energy Cost prediction should include both CCO and enhanced traffic distribution offloading action to improve energy efficiency.

[0552] Proactive Energy Saving Strategies:

[0553] Traditional networks often react to traffic changes after they occur. However, predictive analytics and proactive measures can significantly enhance energy efficiency.

[0554] The EC prediction function, when informed by CCO and mobility data, can forecast potential traffic patterns and energy consumption trends. This predictive capability enables the network to pre-emptively adjust parameters, such as reducing power in areas with anticipated low demand or preparing cells for higher traffic. This proactive approach helps in smoothing out energy consumption peaks and valleys, resulting in more consistent and lower overall energy usage.

[0555] An Energy Cost prediction with proactive energy saving strategy would enhance energy efficiency.

[0556] A proper algorithm effectively integrates CCO with the potential offloading of traffic to neighboring cells into an EC prediction framework for a single gNB. By dynamically adjusting transmission power, optimizing beamforming, leveraging neighboring cells' capacity, and employing sleep modes, the algorithm significantly reduces energy consumption. The continuous learning and feedback loop ensure that the system adapts and optimizes performance, achieving the ultimate goal of enhanced energy savings.

[0557] During the initialization process, a set of input parameters and output parameters may be defined. For example, the input parameters may include a maximum transmission power, a minimum transmission power, a power consumption in sleep mode, a traffic load threshold for activating sleep mode, a minimum SSB beam shaping angle width, a maximum capacity threshold, neighbor cells data (capabilities, current load, and potential for handling offloaded traffic), historical data for training AI / ML models. In another example, the output parameters may include a baseline energy cost and an optimized energy cost.

[0558] The method may involve performing real-time data collection, which includes the systematic gathering of Real-Time Metrics. In this context, the real-time metrics may encompass various network performance indicators, such as the Synchronization Signal Signal-to- Interference-plus-Noise Ratio (SS-SINR), the Synchronization Signal Reference Signal Received Power (SS-RSRP), and the Synchronization Signal Reference Signal Received Quality (SS- RSRQ). Additionally, the real-time metrics may include various traffic metrics having details of the current traffic load, the number of Radio Resource Control (RRC) connected user equipment (UE), and the distribution of UEs across the network. Furthermore, the real-time metrics may include neighboring cells' load and capacity, channel prediction metrics having channel modeling predictions that account for physical obstructions, allowing for enhanced accuracy in forecasting channel conditions and improving overall network performance.

[0559] A baseline energy cost is calculated. In an example, the baseline energy cost refers to the standard or reference level of energy consumption associated with the operation of a system or network under typical conditions. In an example, the baseline energy cost may be calculated using the following Equation:based on current power consumption without CCO optimizations, where PCurrent(. ) is the power consumption at time t during the time interval of measurements and dt.

[0560] The method may involve AI / ML-based prediction, encompassing wo key areas: traffic and load forecasting, and coverage demand analysis. For example, in the traffic and load forecasting, one or more Machine learning (AI / ML) models may be employed to analyze historical and real-time data in order to predict future traffic patterns and load distribution across the network. This predictive capability of the AI / ML models enables the system to anticipate changes in user demand, helping to optimize resource allocation and enhance network performance. In an aspect, the AI / ML models may be able to perform coverage demand analysis. The AI / ML models focus on predicting coverage demand, user mobility patterns, and user distribution (density) in coverage demand analysis. The AI / ML models may perform neighboring cell assessment. For example, AI / ML models may evaluate neighboring cells' current and projected load capacity for potential offloading. By leveraging the AI / ML models, the system can forecast areas of high user concentration and anticipated movement, allowing for proactive adjustments in network configuration and resource management to ensure adequate coverage and seamless connectivity.

[0561] The method may involve dynamic adjustment of network parameters. In an aspect, for dynamic adjustment of network parameters, the present disclosure may employ adaptive transmission power control, beamforming optimization, and sleep mode activation. The adaptive transmission power control enables the network to adjust transmission power levels in real time based on the predicted traffic patterns, load distribution, and coverage demand identified inprevious steps. By optimizing transmission power dynamically, the network can enhance signal quality for users while minimizing energy consumption and reducing interference with neighboring cells. This adaptive approach ensures that power is allocated efficiently, allowing for improved overall network performance and user experience, especially in varying conditions such as fluctuating traffic loads and user mobility patterns. In the adaptive transmission power control, the present disclosure may calculate an optimal transmission power using following equation:

[0562] The optimal transmission power may be calculated based on predicted traffic load and coverage requirements, where Lpred(t) is the predicted traffic load at time t.

[0563] In the beamforming optimization, the present disclosure may adjust beamforming angles and widths based on user density and demand using the following equation.In an aspect, the UEpred(t)' is the predicted UE distribution.

[0564] In the sleep mode activation, the present disclosure may be configured to determine If Lcurrent< Lthand activate a transition of gNB (network node) to sleep mode. In the sleep mode, the power may be defined as

[0565] According to an implementation, traffic offloading to neighboring cells may be performed based on offloading criteria. In an implementation, traffic offloading may be triggered when current traffic load is greater than the traffic load threshold for activating sleep mode (when ^current > Lth)orduring predicted high-demand periods.

[0566] In an implementation, target cells may be selected from amongst the neighboring cells. For example, neighboring cells may be selected based on their current and predicted load, coverage overlap, and capacity. Target cells may be determined based on the following Equation.

[0567] In an implementation, offloading execution may be performed to handover UEs to selected neighboring cells to balance load and reduce energy consumption. In an implementation, the offloading may be performed using the Equation below.

[0568] The method may involve real-time energy cost (EC) calculation and optimization. In this step, the present disclosure continuously monitors and calculates the current EC of the network based on the dynamically adjusted parameters and operational states. In an aspect, the present disclosure may calculate optimized EC using the following equation.

[0569] In an aspect, the present disclosure may be further configured to compare the calculated Optimized EC with the baseline EC. Based on the comparison, the present disclosure may adjust offloading thresholds, power levels and beamforming optimization parameters if ECoptis not sufficiently lower than ECbase.

[0570] The method may involve continuous learning and feedback loop. In an aspect, the present disclosure may opt for a performance evaluation, a model refinement, and an optimization refinement for continuous learning and a feedback loop. In the performance evaluation, the present disclosure may evaluate the performance of optimizations in terms of energy savings and network quality. In the model refinement, the present disclosure may update AI / ML models with new data for improved prediction accuracy. In the optimization refinement, the present disclosure may refine CCO strategies, offloading strategies, and parameters based on performance evaluations and feedback.

[0571] In an implementation, a proper algorithm effectively integrates CCO with the potential offloading of traffic to neighboring cells into an EC prediction framework for a single gNB. By dynamically adjusting transmission power, optimizing beamforming, leveraging neighboring cells' capacity, and employing sleep modes, the algorithm significantly reduces energy consumption. The continuous learning and feedback loop ensures that the system adapts and optimizes performance, achieving the ultimate goal of enhanced energy savings in 5G RAN.

[0572] In an embodiment, an Energy Cost prediction should be derived with a mutual exclusive consideration of a potential offloading action in coordination with CCO beam improvements. Thus, a holistic approach of EC function would enhance the energy saving actions between NG-RAN nodes in order to be interpretable or actionable by the receiving node.

[0573] The intricate interplay between offloading and CCO beam improvements can significantly enhance energy efficiency in NG-RAN networks. Network operators can achieve substantial energy savings by strategically offloading UEs with high EC profiles and optimizing beam configurations. The decoupling high-energy demand nodes from those capable of efficient energy management creates a more balanced and efficient network. This approach not only reduces overall energy costs but also improves the sustainability and reliability of the network, aligningwith the broader goals of modern telecommunications to support growing data demands while minimizing environmental impact.

[0574] Consequently, a predicted high EC might result from the NG-RAN node serving several UEs with low, bursty, but frequent traffic that prevents it from enabling energy saving features due to offloading, but might be assisted by the energy saving actions due to CCO beam improvements. Hence, the neighboring NG-RAN node may be able to serve the new, offloaded UEs with a marginal increment in its EC; at the same time, the offloading might enable the source NG-RAN node to reduce its own EC significantly and to optimize its energy saving by proper CCO optimization actions.

[0575] In an embodiment, an Energy Cost prediction on one node should be tolerated with respect to a potential offloading action between neighbor NG-RAN nodes with potential CCO beam optimization action capabilities.

[0576] Under this paradigm, the source node of the offloading action should signal to the target node a description of the “additional load” for the target node; it would then receive from the target node a prediction of the Energy Cost at the target node, assuming the offloading action was successfully executed.

[0577] Follows a proposed algorithm to integrate Coverage and Capacity Optimization (CCO) with offloading actions into an Energy Cost (EC) prediction function for a gNB in a 5G RAN in a more flexible mutually exclusive way. The primary goal is to enhance energy savings by optimizing network performance and resource usage. The algorithm follows specific rules: offloading conditions are mutually exclusive to CCO actions, offloading should include CCO considerations, and CCO actions should be prioritized to manage the load before considering offloading.

[0578] 1. Initialization• Input Parameters: o Pmax'- Maximum transmission power. o Pmin- Minimum transmission power. o Psleep- Power consumption in sleep mode. o Lth ccoTraffic CCO load activation threshold. o Lth_offioad'- Traffic offload (mobility) activation threshold o BminMinimum SSB beam shaping angle width. o Neighbor Cells Data: current load and capacity info.o Historical traffic profile data: data to understand traffic patterns based on AI / ML models.• Output Parameters: o ECbaseBaseline Energy Cost. o ECoptOptimized Energy Cost.

[0579] 2. Real-Time Data Collection• Collect Real-Time Metrics: o Network performance metrics: SS-SINR, SS-RSRP, SS-RSRQ. o Traffic metrics: Current traffic load (RRC Connected UEs), user equipment (UE) distribution. o Channel prediction: Channel modeling predictions based on physical obstructions. o Neighboring cells' load and capacityBaseline Energy Cost Calculation o Calculate:based on current power consumption without CCO optimizations, where PCurrent(E) is the power consumption at time t during the time interval of measurements and dt.

[0580] 3. AI / ML-Based Prediction• Traffic and Load Forecasting: o Use machine learning models to predict future traffic patterns and load distribution based on historical and real-time data.• Coverage and Capacity Demand Analysis: o Predict coverage and capacity demand and user mobility patterns and user distribution (density).• Neighbor Cell Assessment: o Evaluate neighboring cells' current and projected load capacity for potential offloading.

[0581] 4. Dynamic Adjustment of Network ParametersA. Check current load against thresholds• Low Load Conditions: o ifno additional action is required, with potential sleep mode if low traffic conditions persist• Intermediate Load Conditions: o if additional action is required with CCOactivation.B. CCO actions consideration• Power control optimization: o Calculate optimal transmission power as:based on predicted traffic load and coverage requirements, where Lpred(t) is the predicted traffic load at time t.Beamforming adjustments: o Adjust beamforming angles and widths based on user density and demand to focus coverage and minimize interference:where UEpred) is the predicted UE distribution.• Evaluate impact on EC function: o Calculate the potential reduction in EC due to CCO:• Check sufficiency of CCO actions: o Assess if CCO actions sufficiently reduce load and EC without needing offloadingC. Traffic offloading Consideration to neighbor cells• Offloading Criteria: o Consider potential offloading when Lcurrent> Lthor during predicted high- demand periods.D. Traffic offloading Actions to neighbor cells• Selection of Target cells: o Select neighboring cells based on their current and predicted load, coverage overlap, and capacity:• Offloading execution: o Handover UEs to selected neighboring cells to balance load and reduce energy consumption:• Combined CCO and offloading EC calculation: o Calculate the total EC with offloading and additional CCO:• Criteria for Sleep Mode: o If Lcurrentis significantly reduced post offloading, consider transitioning gNB to sleep mode:

[0582] 5. Real-Time Energy Cost Calculation and Optimization• Comparison and Adjustment: o Continuously compare ECopt, ECbaseto assess energy savings o Adjust offloading thresholds, power levels and beamforming optimization parameters if ECoptis not sufficiently lower than ECbase.• Performance Evaluation: o Evaluate the performance of optimizations in terms of energy savings and network quality.• Model Refinement: o Update AI / ML models with new data for improved prediction accuracy.• Optimization Refinement: o Refine CCO and offloading strategies and parameters based on performance evaluations and feedback.

[0583] Furthermore, in an embodiment, the potential load may be expanded to contain more parameters and metrics / variables than the number of UEs. The EC load considers multiple factors as follows:1. Traffic Characteristics: Differentiating between traffic types and their respective bandwidth, latency, and QoS requirements. Hence, other factors to contribute to the EC function might be the PRB utilization, UL / DL data volume, etc.2. Radio and Environmental Conditions: Accounting for variations in signal quality (UE RSRP,SINR) and the additional resources (i.e., RSRP / SINR vs. PRB) needed for UEs in challenging conditions.3. Temporal Variations: Recognizing patterns over time, such as peak hours or event-driven traffic spikes. This could include RRM measurements.4. Behavioral Insights: Understanding user mobility patterns and their impact on resource usage based on QoS characteristics.

[0584] The metrics may be selected by considering the impact on the energy consumption the UE may generate, such as UE capabilities and types of services used.

[0585] In an embodiment, incorporating a wide range of detailed metrics and factors may enhance the analysis, though it may increase processing requirements and complexity. By thoughtfully selecting the EC and relevant metrics, performance is optimized while maintaining manageable processing loads and allowing for future enhancements.

[0586] FIG. 4 illustrates another exemplary flow diagram (400) of a method for energy cost (EC) optimization based on one or more beamforming actions in the network (108), in accordance with an embodiment of the present disclosure.

[0587] At step (402), the method (400) includes receiving, by the receiving unit (118), a plurality of first parameters. The plurality of first parameters comprises, but is not limited to, maximum transmission power, minimum transmission power, power consumption in a sleep mode, a traffic load threshold for activating sleep mode, minimum synchronization signal block (SSB) beam shaping angle width, maximum capacity threshold and historical data for training the one or more ML models. The plurality of first parameters is provided by a network operator. The network operator provides the plurality of first parameters based on network requirements, for example, usage traffic, and performance metrics.

[0588] At step (404), the method (400) includes collecting, by the collection unit (128), one or more real-time metrics from the network. The one or more real-time metrics comprise one or more network performance metrics, one or more traffic metrics and channel predictions. The one or more network performance metrics comprise a synchronization signal (SS)- signal to interference plus noise ratio (SINR), a SS-reference signal received power (SS-RSRP), a SS- reference signal received quality (SS-RSRQ). The one or more traffic metrics comprise traffic load at the time t, user equipment (UE) distribution, counters and key performance indicators (KPIs). The channel prediction comprises channel modeling predictions based on physical obstructions, to be used for beam shaping, beam-steering, beam-control and beam-management actions.

[0589] At step (406), the method (400) includes calculating, by the processing unit (120), a baseline EC based on power consumption at time t. The baseline EC is calculated as

[0590] The PCurrent( ) is the power consumption between a time interval from the time t to time T, and At is a time interval between measurements. In an aspect, the baseline EC provides the expected or reference energy consumption and cost under normal or standard operating conditions, without any interventions, changes, or optimizations. The baseline EC serves as a benchmark to compare actual or optimized energy usage against. This enables evaluation of energy savings, efficiency measures, or cost impacts of changes in operations, equipment, or behavior.

[0591] At step (408), the method (400) includes performing, by the processing unit (120), one or more analysis types, to predict a plurality of second parameters, using one or more machine learning (ML) models based on the plurality of first parameters and the one or more collected realtime metrics. The one or more analysis type comprises a traffic and load forecasting analysis, and a coverage demand analysis. The plurality of second parameters comprises one or more traffic patterns, a traffic load, a coverage demand, and one or more user mobility patterns.

[0592] In an aspect, the traffic and load forecasting analysis refer to a process of predicting the one or more traffic patterns and traffic load based on user behavior and network usage. The traffic patterns comprise, for example, traffic behavior (e.g., high traffic, moderate traffic, low traffic, etc.), spatial or geographical traffic pattern (e.g., urban, rural, suburban, event, stadium, highway, etc.), seasonal (e.g., weekday, weekend, holiday, summer, vacation), app-based (e.g., social, streaming, work apps, gaming, etc.). The traffic load is predicted based on user activity, weather, and mobility patterns. The traffic load may comprise, but is not limited to, voice traffic, data traffic, app traffic, control signaling traffic, synchronization traffic, security traffic and authentication traffic.

[0593] In an aspect, the coverage demand analysis refers to a process of predicting the coverage demand and the user mobility patterns based on user density, terrain, building structures, and usage behavior. The coverage and capacity demand analysis is used to evaluate coverage demand (i.e., whether network signals reliably reach the geographical areas where users need service, with sufficient signal strength and quality). The coverage and capacity analysis is used to evaluate the user mobility patterns. The user mobility patterns refer to movement behaviors and trajectories of users as the users travel through different geographical areas covered by the network. The mobility patterns comprise movement paths (e.g., highways, local roads), speed and direction (walking, driving, stationary).

[0594] At step (410), the method (400) includes, upon performing, executing, by the processing unit (120), the one or more beamforming actions based on the plurality of first parameters and at least one second parameter of the plurality of second parameters. The one or more beamforming actions comprise beam shaping, beam steering, beam control and beam management.

[0595] In an aspect, the execution of the one or more beamforming actions comprises determining of a beamforming gain and a beamwidth based on a number of active antenna sub-arrays. Total energy consumption for the number of active antenna sub-arrays is determined based on one or more antenna parameters. The one or more antenna parameters comprise the number of active sub-arrays, a transmit power per sub-array, a baseband processing power per active radio frequency (RF) chain, a power amplifier consumption, and a minimum beamforming gain.

[0596] A normalized cost function is determined based on the beamforming gain, the total energy consumption, and the beamwidth. In an aspect, the normalized cost function refers to a cost function that has been scaled (or normalized) so that its output values fall within a standardized range (e.g., between 0 and 1). In an operative aspect, the normalized cost function is determined based on the beamforming gain, total energy consumption and the beamwidth. Lower normalized cost (i.e., close to 0) is better, indicating high beamforming gain, low energy consumption, and high bandwidth. Higher normalized cost (i.e., close to 1) indicates poor performance. The normalized cost function is determined by considering constraints such as the beamforming gain constrainttotal transmit power constraint and minimumbeamwidth coverage constraint

[0597] An optimization loop algorithm is applied based on the minimum beamforming gain, the maximum transmission power, a target beamwidth, and a total number of sub-arrays. In an aspect, the optimization loop algorithm is a loop-based process that repeatedly adjusts parameters (e.g., beamforming angle, transmit power, or array configuration) to maximize signal strength, minimize interference and ensure coverage meets user demands

[0598] In the optimization loop algorithm, the beamforming gain, the minimum beamwidth, and the transmit power are calculated for each active sub-array. The total energy consumption is calculated for each active sub-array. Further, a cost function for each active array is calculated based on the number of active sub-arrays and the total energy consumption. A minimum cost function is selected from each estimated cost function corresponding to the number of active sub-arrays. The beam configuration of sub-array having minimum cost function is considered as an optimal energy-optimized beam configuration. The optimal energy-optimized beam configuration is applied on the sub-arrays.

[0599] At step (412), the method (400) includes activating, by the processing unit (120), a sleep mode based on the one or more performed beamforming actions. The sleep mode activation comprises monitoring of a real-time cell load. The real-time cell load is compared with a load threshold to determine whether the real-time cell load is less than the load threshold. Upon determining that the real-time cell load is less than the load threshold, the sleep mode is executed on the active sub-arrays. Further, in the sleep mode execution, the sleep mode is activated on the number of active sub-arrays by setting the number of active sub-arrays as number of sleep subarrays. After activating the sleep mode on the number of sleep sub-arrays, the real-time cell load is again monitored after the predefined time interval (e.g., 5 mins set by the network operator based on network requirements) to determine whether the real-time cell load is greater than the loadthreshold. Upon determining that the real-time cell load is greater than the load threshold, a normal mode is activated on the number of active sub-arrays.

[0600] At step (414), the method (400) includes calculating, by the processing unit (120), an optimized EC based on the plurality of first parameters and the plurality of second parameters. In an operative aspect, the optimized EC is calculated based on the maximum transmission power, the minimum transmission power and the predicted load at the time t. The optimized EC provides energy consumption value obtained after applying optimization techniques to minimize power usage while maintaining desired network performance.

[0601] At step (416), the method (400) includes comparing, by the determining unit (122), the optimized EC and the baseline EC. The optimized energy cost (EC) is compared with the baseline EC to validate that optimization reduces energy consumption while satisfying the constraints (e.g., beamforming gain, total transmit power, and minimum beamwidth).

[0602] At step (418), the method (400) includes upon determining that the optimized EC is not less than the baseline EC by a predefined margin, re-executing, by the processing unit (120), the one or more beamforming actions to adjust one or more beamforming parameters. In an aspect, the optimized EC (ECopt) is compared with the baseline EC (ECbase) to determine whether the optimized EC is not less than the baseline EC by the predefined margin which is a set threshold or minimum allowable difference that quantifies the level of improvement required for the optimization to be considered significant or effective. The optimized EC is not less than the baseline EC by the predefined margin, so the optimization is not accepted. When the optimized EC is less than the baseline EC by the predefined margin, the optimization is accepted. This achieves energy savings.

[0603] The re-execution of the one or more beamforming actions comprises adjusting, by the processing unit (120), one or more beamforming parameters and the one or more ML models. The one or more beamforming parameters comprise SSB beam shaping angle width. In an aspect, the beamforming parameters (i.e., beam-shaping, beam-steering) are adjusted to fulfill the condition ECopt«ECbaseat the time period T. Further, the continuous learning and feedback loop is carried out to perform processes, for example, performance assessment, model updating, and optimization tuning. During the performance assessment, the effectiveness of the optimizations is measured with respect to energy efficiency and overall network quality. In the model updating phase, AI / ML models are retrained or adjusted using fresh data to enhance the prediction capabilities. Finally, in the optimization tuning step, the CCO strategies and parameters are adjusted and improved based on the results of performance assessments and received feedback.

[0604] In an embodiment, a method for energy cost (EC) optimization based on one or more beamforming actions in a network. The method comprises initialization by defining input parameters and output parameters. The input parameters comprise, but are not limited to, maximum transmission power, minimum transmission power, power consumption in a sleepmode, traffic load threshold for activating the sleep mode, minimum synchronization signal block (SSB) beam shaping angle width, maximum capacity threshold and historical data for training artificial intelligence / machine learning (AI / ML). The output parameters comprise, but are not limited to, baseline energy cost and optimized energy cost. The method includes performing realtime data collection, systematically gathering real-time metrics. The real-time metrics comprises, but are not limited to, network performance SSB metrics, traffic metrics and channel predictions. The method comprises calculating the baseline energy cost based on power consumption at time t (i.e., current time t). The method comprises conducting AI / ML-based prediction for traffic and load forecasting and coverage demand analysis, using historical and real-time data to predict future patterns and traffic load and allowing for resource optimization and proactive adjustments. The method may include dynamically adjusting network parameters (i.e., beamforming parameters such as beam angle and width) by performing one or more beamforming actions, for example, beam shaping, beam steering, beam control and beam management. Further, the method comprises calculating of the optimized EC based on the input parameters and the predicted traffic load. Realtime energy cost (EC) is optimized by continuously monitoring the network's current EC based on dynamically adjusted parameters. The optimized EC is compared with the baseline EC to adjust network parameters as needed. Further, the network performance of optimization is continuously evaluated in terms of energy savings and network quality. The AI / ML models are updated with new data to improve prediction accuracy. CCO strategies and parameters are refined based on performance evaluation and feedback.

[0605] FIG. 5 illustrates an exemplary computer system (500) in which or with which embodiments of the present disclosure may be implemented.

[0606] As shown in FIG. 5, the computer system (500) may include an external storage device (510), a bus (520), a main memory (530), a read-only memory (540), a mass storage device (550), communication port(s) (560), and a processor (570). A person skilled in the art will appreciate that the computer system may include more than one processor and communication ports. The processor (570) may include various modules associated with embodiments of the present disclosure. The communication port(s) (560) may be any of an RS-232 port for use with a modem-based dialup connection, a 10 / 100 Ethernet port, a Gigabit or 10 Gigabit port using copper or fiber, a serial port, a parallel port, or other existing or future ports. The communication port(s) (560) may be chosen depending on a network, such a Local Area Network (LAN), Wide Area Network (WAN), or any network to which the computer system connects.

[0607] The main memory (530) may be random access memory (RAM), or any other dynamic storage device commonly known in the art. The read-only memory (540) may be any static storage device(s) e.g., but not limited to, a Programmable Read Only Memory (PROM) chips for storing static information e.g., start-up or Basic Input / Output System (BIOS) instructions for the processor (570). The mass storage device (550) may be any current or future mass storage solution, which can be used to store information and / or instructions. Exemplary mass storage device (550) includes, but is not limited to, Parallel Advanced Technology Attachment (PAT A) orSerial Advanced Technology Attachment (SATA) hard disk drives or solid-state drives (internal or external, e.g., having Universal Serial Bus (USB) and / or Firewire interfaces), one or more optical discs, Redundant Array of Independent Disks (RAID) storage, e.g., an array of disks.

[0608] The bus (520) communicatively couples the processor (570) with the other memory, storage, and communication blocks. The bus (520) may be, e.g., a Peripheral Component Interconnect (PCI) / PCI Extended (PCI-X) bus, Small Computer System Interface (SCSI), Universal Serial Bus (USB), or the like, for connecting expansion cards, drives, and other subsystems as well as other buses, such a front side bus (FSB), which connects the processor (570) to the computer system.

[0609] Optionally, operator and administrative interfaces, e.g., a display, keyboard, joystick, and a cursor control device, may also be coupled to the bus (520) to support direct operator interaction with the computer system. Other operator and administrative interfaces can be provided through network connections connected through the communication port(s) (560). Components described above are meant only to exemplify various possibilities. In no way should the aforementioned exemplary computer system limit the scope of the present disclosure.

[0610] The exemplary computer system (500) is configured to execute a computer program product comprising a non-transitory computer-readable medium comprising instructions that, when executed by one or more processors, cause the one or more processors to perform a method for energy cost (EC) optimization based on one or more beamforming actions in a network is described. The method comprises receiving, by a receiving unit, a plurality of first parameters and collecting, by a collection unit, one or more real-time metrics from the network. The method comprises calculating, by a processing unit, a baseline EC based on power consumption at time t. The method comprises performing, by the processing unit, one or more analysis types, to predict a plurality of second parameters, using one or more machine learning (ML) models based on the plurality of first parameters and the one or more collected real-time metrics. The method comprises upon performing, executing, by the processing unit, the one or more beamforming actions based on the plurality of first parameters and at least one second parameter of the plurality of second parameters. The method comprises activating, by the processing unit, a sleep mode based on the one or more performed beamforming actions. The method comprises calculating, by the processing unit, an optimized EC based on the plurality of first parameters and the plurality of second parameters. The method comprises comparing, by a determining unit, the optimized EC and the baseline EC. The method comprises upon determining that the optimized EC is not less than the baseline EC by a predefined margin, re-executing, by the processing unit, the one or more beamforming actions to adjust one or more beamforming parameters.

[0611] The present disclosure provides technical advancements related to energy management in the network. Advancement addresses the limitations of existing solutions by facilitating energy cost (EC) prediction based on one or more beamforming actions (beam shaping, beam steering, beam control, and beam management) in the network. Multiple SSB beam shapesare dynamically selected with multiple beamwidth configurations and SSB beam power (SSB power boosting). The multiple SSB beam shapes are updated with the multiple beamwidth configurations and the SSB beam power per sector based on an energy cost (EC) function. Beamforming strategies to improve network coverage and network capacity with energy-saving and network load processing. Downlink (DL) coverage and SINR are improved with energy constraints. In this way, a better, more flexible and more adaptive beam management for the overall system performance optimization and energy efficiency sustainability.

[0612] While the foregoing describes various embodiments of the invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof. The scope of the invention is determined by the claims that follow. The invention is not limited to the described embodiments, versions or examples, which are included to enable a person having ordinary skill in the art to make and use the invention when combined with information and knowledge available to the person having ordinary skill in the art.ADVANTEGES OF THE PRESENT DISCLOSURE

[0613] The present disclosure provides a system and a method for energy cost (EC) optimization based on one or more beamforming actions in the network.

[0614] The present disclosure provides a system and a method for adaptive SSB configuration with integrated beam shaping, energy conservation, and offloading.

[0615] The present disclosure performs an SSB beam boosting, an SSB beam shaping, and an energy cost (EC) prediction energy saving with potential offloading.

[0616] The present disclosure predicts energy cost (EC) for energy savings based on coverage and capacity optimization (CCO) with a potential neighboring cell offloading.

[0617] The present disclosure dynamically selects multiple SSB beam shapes with multiple beamwidth configurations and SSB beam power (SSB power boosting).

[0618] The present disclosure updates the multiple SSB beam shapes with the multiple beam width configurations and the SSB beam power per sector based on an energy cost (EC) function.

[0619] The present disclosure dynamically adjusts beamforming strategies to improve network coverage and network capacity with energy-saving and network load processing.

[0620] The present disclosure improves downlink (DL) coverage and SINR with energy constraints.

[0621] The present disclosure provides a better, more flexible and more adaptive Beam management for the overall system performance optimization and energy efficiency sustainability.

[0622] The present disclosure employs a combined algorithmic approach, based on a functional integration of several system functional integrities like SSB beam shaping, SSB beam steering, Energy Cost (EC) function and offloading techniques, which will enhance the system performance and preserve the green energy network retainability.

[0623] The Energy Cost prediction function plays a critical role in managing the energy efficiency of dense networks.

[0624] The SSB multi-beam management allows for the simultaneous control and optimization of multiple beams within a network. The offloading, or handover, actions are critical in maintaining optimal load distribution across the network.

[0625] In dense deployments, efficient handover mechanisms ensure that users are seamlessly transitioned between cells, minimizing congestion and optimizing resource utilization. The Energy Cost prediction function can inform these handover decisions by predicting the energy costs associated with different handover scenarios. Finally, the SSB beam shaping involves adjusting the beam's shape to focus signal energy on specific directions, enhancing coverage and signal quality in targeted areas.

[0626] Azimuth and elevation angle steering is refined by adjusting the direction of the beams in three-dimensional space. These capabilities are particularly important in dense networks, where precise control over beam direction can significantly reduce interference and improve user experience.

[0627] The integration of these technologies and functions offers a comprehensive approach to energy savings and system performance optimization.

[0628] The predictive capabilities of the energy cost prediction function, combined with the dynamic control offered by SSB multi-beam management, beam shaping, and steering, allow for real-time adjustments that balance energy consumption with network performance.

[0629] Offloading mechanisms complement these efforts by efficiently distributing user load, reducing the strain on any single cell and optimizing overall network efficiency.

[0630] The present disclosure is related mainly to gNB CU / DU and / or 0AM functional blocks. In an embodiment, the present disclosure addresses SSB (Synchronization Signal Block) multi-beam management in dense 5G and upcoming 6G networks.

[0631] The present disclosure integrates energy cost prediction, power control, offloading (handover) actions, beam shaping, and beam steering (azimuth / elevation).

[0632] The present disclosure improves network coverage and capacity with energysaving and network load processing.

Claims

CLAIMSWe claim:

1. A method (400) for energy cost (EC) optimization based on one or more beamforming actions in a network (108), the method (400) comprising: receiving (402), by a receiving unit (118), a plurality of first parameters; collecting (404), by a collection unit (128), one or more real-time metrics from the network; calculating (406), by a processing unit (120), a baseline EC based on power consumption at time t; performing (408), by the processing unit (120), one or more analysis types, to predict a plurality of second parameters, using one or more machine learning (ML) models based on the plurality of first parameters and the one or more collected real-time metrics; upon performing, executing (410), by the processing unit (120), the one or more beamforming actions based on the plurality of first parameters and at least one second parameter of the plurality of second parameters; activating (412), by the processing unit (120), a sleep mode based on the one or more performed beamforming actions; calculating (414), by the processing unit (120), an optimized EC based on the plurality of first parameters and the plurality of second parameters; comparing (416), by a determining unit (122), the optimized EC and the baseline EC; and upon determining that the optimized EC is not less than the baseline EC by a predefined margin, re-executing (418), by the processing unit (120), the one or more beamforming actions to adjust one or more beamforming parameters.

2. The method (400) as claimed in claim 1, wherein the plurality of first parameters comprises maximum transmission power, minimum transmission power, power consumption in a sleep mode, a traffic load threshold for activating sleep mode, minimum synchronization signal block (SSB) beam shaping angle width, maximum capacity threshold and historical data for training the one or more ML models, wherein the plurality of first parameters is provided by a network operator.

3. The method (400) as claimed in claim 1 , wherein the one or more real-time metrics comprise one or more network performance metrics, one or more traffic metrics and channelpredictions, wherein the one or more network performance metrics comprise a synchronization signal (SS)- signal to interference plus noise ratio (SINR), a SS -reference signal received power (SS-RSRP), a SS-reference signal received quality (SS-RSRQ), wherein the one or more traffic metrics comprise traffic load at the time t, user equipment (UE) distribution, counters and key performance indicators (KPIs).

4. The method (400) as claimed in claim 1, wherein the baseline EC is calculated aswherein the Pcurrent ) is the power consumption between a time interval from the time t to time T, and is a time interval between measurements.

5. The method (400) as claimed in claim 1, wherein the one or more analysis type comprises a traffic and load forecasting analysis, and a coverage demand analysis.

6. The method (400) as claimed in claim 1, wherein the plurality of second parameters comprises one or more traffic patterns, a traffic load, a coverage demand, and one or more user mobility patterns.

7. The method (400) as claimed in claim 1, wherein the one or more beamforming actions comprise beam shaping, beam steering, beam control and beam management.

8. The method (400) as claimed in claim 1, wherein the execution of the one or more beamforming actions comprising: determining, by the determining unit (122), a beamforming gain and a beam width based on a number of active antenna sub-arrays; determining, by the determining unit (122), a total energy consumption for the number of active antenna sub-arrays based on one or more antenna parameters, wherein the one or more antenna parameters comprise the number of active sub-arrays, a transmit power per sub-array, a baseband processing power per active radio frequency (RF) chain, a power amplifier consumption, and a minimum beamforming gain; determining, by the determining unit (122), a normalized cost function based on the beamforming gain, the total energy consumption, and the beamwidth; and applying, by the processing unit (120), an optimization loop algorithm based on the minimum beamforming gain, the maximum transmission power, a target beamwidth, and a total number of sub-arrays.

9. The method (400) as claimed in claim 8, wherein the optimization loop algorithm comprising:calculating, by the processing unit (120), the beamforming gain, the minimum beamwidth, and the transmit power for each active sub-array; calculating, by the processing unit (120), the total energy consumption for each active sub-array; estimating, by the processing unit (120), a cost function for each active array based on the number of active sub-arrays and the total energy consumption; selecting, by the processing unit (120), a minimum cost function from each estimated cost function corresponding to the number of active sub-arrays; and outputting, by the processing unit (120), an optimal energy-optimized beam configuration.

10. The method (400) as claimed in claim 1, wherein the sleep mode activation comprising: monitoring, by the processing unit (120), a real-time cell load; determining, by the determining unit (122), whether the real-time cell load is less than a load threshold; and upon determining that the real-time cell load is less than the load threshold, executing, by the processing unit (120), the sleep mode, wherein the sleep mode execution comprising: activating, by the processing unit (120), the sleep mode on the number of active sub-arrays; determining, by the determining unit (122), whether the real-time cell load is greater than the load threshold after a predefined time interval; and upon determining that the real-time cell load is greater than the load threshold, activating, by the processing unit (120), a normal mode on the number of active sub-arrays.

11. The method (400) as claimed in claim 1, wherein the re-execution of the one or more beamforming actions comprising: adjusting, by the processing unit (120), one or more beamforming parameters and the one or more ML models, wherein the one or more beamforming parameters comprise SSB beam shaping angle width.

12. A system (106) for energy cost (EC) optimization based on one or more beamforming actions in a network (108), the system (106) comprising: a receiving unit (118) configured to receive a plurality of first parameters;a collection unit (128) configured to collect one or more real-time metrics from the network; a processing unit (120) configured to: calculate a baseline EC based on power consumption at time t; perform one or more analysis types, to predict a plurality of second parameters, using one or more machine learning (ML) models based on the plurality of first parameters and the one or more collected real-time metrics; upon performing, execute the one or more beamforming actions based on the plurality of first parameters and at least one second parameter of the plurality of second parameters; activate a sleep mode based on the one or more performed beamforming actions; and calculate an optimized EC based on the plurality of first parameters and the plurality of second parameters; and a determining unit (122) configured to: compare the optimized EC and the baseline EC; and upon determining that the optimized EC is not less than the baseline EC by a predefined margin, the processing unit (120) configured to re-execute the one or more beamforming actions to adjust one or more beamforming parameters.

13. The system (106) as claimed in claim 12, wherein the plurality of first parameters comprises maximum transmission power, minimum transmission power, power consumption in a sleep mode, a traffic load threshold for activating sleep mode, minimum synchronization signal block (SSB) beam shaping angle width, maximum capacity threshold and historical data for training the one or more ML models, wherein the plurality of first parameters are provided by a network operator.

14. The system (106) as claimed in claim 12, wherein the one or more real-time metrics comprise one or more network performance metrics, one or more traffic metrics and channel predictions, wherein the one or more network performance metrics comprise a synchronization signal (SS)- signal to interference plus noise ratio (SINR), a SS -reference signal received power (SS-RSRP), a SS-reference signal received quality (SS-RSRQ), wherein the one or more traffic metrics comprise traffic load at the time t, user equipment (UE) distribution, counters and key performance indicators (KPIs).

15. The system (106) as claimed in claim 12, wherein the baseline EC is calculated aswherein the Pcurrent(t) is the power consumptionbetween a time interval from the time t to time T, and At is a time interval between measurements.

16. The system (106) as claimed in claim 12, wherein the one or more analysis type comprises a traffic and load forecasting analysis, and a coverage demand analysis.

17. The system (106) as claimed in claim 12, wherein the plurality of second parameters comprises one or more traffic patterns, a traffic load, a coverage demand, and one or more user mobility patterns.

18. The system (106) as claimed in claim 12, wherein the one or more beamforming actions comprise a beam shaping, a beam steering, a beam control and a beam management.

19. The system (106) as claimed in claim 12, wherein the execution of the one or more beamforming actions comprising: the determining unit (122) is configured to: determine a beamforming gain and a beamwidth based on a number of active antenna sub-arrays; determine a total energy consumption for the number of active antenna sub- arrays based on one or more antenna parameters, wherein the one or more antenna parameters comprise the number of active sub-arrays, a transmit power per subarray, a baseband processing power per active radio frequency (RF) chain, a power amplifier consumption, and a minimum beamforming gain; and determine a normalized cost function based on the beamforming gain, the total energy consumption, and the beamwidth; and the processing unit (120) is configured to: apply an optimization loop algorithm based on the minimum beamforming gain, the maximum transmission power, a target beamwidth, and a total number of sub-arrays.

20. The system (106) as claimed in claim 19, wherein the optimization loop algorithm comprising: the processing unit (120) is configured to:calculate the beamforming gain, the minimum beamwidth, and the transmit power for each active sub-array; calculate the total energy consumption for each active sub-array; estimate a cost function for each active array based on the number of active sub-arrays and the total energy consumption; select a minimum cost function from each estimated cost function corresponding to the number of active sub-arrays; and output an optimal energy-optimized beam configuration.

21. The system (106) as claimed in claim 12, wherein the sleep mode activation comprising: the processing unit (120) configured to: monitor a real-time cell load; the determining unit (122) configured to: determine whether the real-time cell load is less than a load threshold; and upon determining that the real-time cell load is less than the load threshold, the processing unit (120) is configured to: execute the sleep mode, wherein the sleep mode execution comprising: the processing unit (120) is configured to activate the sleep mode on the number of active sub-arrays; the determining unit (122) is configured to determine whether the real-time cell load is greater than the load threshold after a predefined time interval; and upon determining that the real-time cell load is greater than the load threshold, the processing unit (120) is configured to activate a normal mode on the number of active sub-arrays.

22. The system (106) as claimed in claim 12, wherein the re-execution of the one or more beamforming actions comprising: the processing unit (120) is configured to adjust one or more beamforming parameters and the one or more ML models, wherein the one or more beamforming parameters comprise SSB beam shaping angle width23. A computer program product comprising a non-transitory computer-readable medium comprising instructions that, when executed by one or more processors, cause the one ormore processors to execute a method (400) for energy cost (EC) optimization based on one or more beamforming actions in a network (108), the method (400) comprising: receiving (402), by a receiving unit (118), a plurality of first parameters; collecting (404), by a collection unit (128), one or more real-time metrics from the network; calculating (406), by a processing unit (120), a baseline EC based on power consumption at time t; performing (408), by the processing unit (120), one or more analysis types, to predict a plurality of second parameters, using one or more machine learning (ML) models based on the plurality of first parameters and the one or more collected real-time metrics; upon performing, executing (410), by the processing unit (120), the one or more beamforming actions based on the plurality of first parameters and at least one second parameter of the plurality of second parameters; activating (412), by the processing unit (120), a sleep mode based on the one or more performed beamforming actions; calculating (414), by the processing unit (120), an optimized EC based on the plurality of first parameters and the plurality of second parameters; comparing (416), by a determining unit (122), the optimized EC and the baseline EC; and upon determining that the optimized EC is not less than the baseline EC by a predefined margin, re-executing (418), by the processing unit (120), the one or more beamforming actions to adjust one or more beamforming parameters.

Images