Promoting energy-aware admission control with dynamic load balancing in advanced communication networks

Abstract

Provided herein is facilitating energy-aware multi-cell admission control with dynamic load balancing in advanced communication networks. One method includes facilitating, by a system comprising a processor, energy-efficiency aware load balancing for served user equipment. The method also includes facilitating, by the system, control of admission of other user equipment to a communication network. The communication network can be deployed as a disaggregated architecture comprising a central unit, a distributed unit, and a near-real-time radio access network intelligent controller. The cell group can be configured to operate according to a new wireless network communication protocol.

Description

Related applications

[0001] This application claims priority to U.S. nonprovisional patent application No. 18 / 524,736, filed November 30, 2023, entitled “FACILITATING ENERGY AWAREADMISSION CONTROL WITH DYNAMIC LOAD BALANCING IN ADVANCED COMMUNICATION NETWORKS”, the entire contents of which are incorporated herein by reference. Background Technology

[0002] Computing devices are ubiquitous. Given the explosive demand for mobile networks and the emergence of advanced use cases such as streaming and gaming, these networks consume significantly more power than LTE networks. This power consumption can be attributed to the exponential growth in network traffic flowing through these advanced networks and the need for faster processing of complex tasks. Therefore, unique challenges exist regarding network efficiency, especially given the upcoming fifth-generation (5G), new radio (NR), sixth-generation (6G), or other next-generation network communication standards.

[0003] The context of communication networks described above is intended only to provide an overview of the current technology and is not intended to be exhaustive. Further contextual descriptions and the corresponding benefits of some of the various non-limiting embodiments described herein will become more apparent upon reading the following detailed description. Summary of the Invention

[0004] The following presents a simplified overview of the disclosed subject matter to provide a basic understanding of some aspects of the various embodiments. This invention is not a broad summary of the various embodiments. It is neither intended to identify key or essential elements of the various embodiments, nor to describe the scope of the various embodiments. Its sole purpose is to present some concepts of this disclosure in a simplified form as a prelude to the more detailed description that follows.

[0005] The embodiments relate to a method that includes facilitating energy efficiency-aware load balancing of served user equipment by a system including a processor. Energy efficiency-aware load balancing distributes the served user equipment across a cluster of cells in a communication network. The method also includes facilitating control over admission to the communication network by the system. The control facilitating energy efficiency-aware load balancing and admission may include evaluating feedback data representing near real-time quality of service performance metrics and controlling energy efficiency-aware load balancing and admission control based on that feedback data. This control can mitigate quality of service degradation for served user equipment. The communication network may be deployed as a decomposed architecture including a central unit, distributed units, and a near real-time radio access network intelligent controller. The cluster of cells may be configured to operate according to a new radio network communication protocol.

[0006] Facilitating energy-efficiency-aware load balancing may include selecting a first cell in the cell group for offloading a first user equipment (UE) from its service to a second cell in the cell group. This cell group is under the control of a near real-time radio access network intelligent controller. Furthermore, facilitating energy-efficiency-aware load balancing may include providing a centralized unit with information indicating an offloading command for the first UE, based on the selection of the first cell, for verification purposes.

[0007] The above implementation may include determining the result of a connection transfer based on changes in network utility, following the completion of the connection transfer at the user device. Furthermore, the system may transmit these changes in network utility for incorporation into the reinforcement learning model.

[0008] In some implementations, controls that mitigate the degradation of service quality for already served user equipment (UFOs) may include controls that minimize the degradation of service quality for already served UFOs. Controls that facilitate admission to other UFOs may include activating an admission control process based on receiving a connection request from a first UFO among the other UFOs. Furthermore, based on the outcome of the admission control process and on the acceptance of admission policies and utility functions, the method may include selectively admitting the first UFO to the cell group. In some implementations, based on the completion of admission for the first UFO, the method may include transmitting cell-level data by the system for incorporation into a reinforcement learning model.

[0009] Another embodiment relates to a system including a processor and a memory storing executable instructions that, when executed by the processor, facilitate the execution of operations. These operations may include, based on the activation of a load balancing process, selecting a first cell among cells for offloading a first user equipment (UE) to a second cell among cells. The cells are under the control of a near real-time radio access network intelligent controller. Furthermore, the load balancing process may include, based on the selection of the first cell, providing information indicative of offloading instructions for the first UE to a centralized unit for verification. The operation may also include, based on receiving a connection request from a second UE, activating an admission control process. Furthermore, based on the outcome of the admission control process, and based on acceptance of an admission policy and utility function, selectively admitting the second UE to a cell. In the example, the load balancing process is activated based on cells within the cluster being determined to meet a utilization threshold. The system may be deployed in a decomposed architecture of network devices.

[0010] In addition to the implementation methods described above, the operation may include determining the result of the connection transfer based on the completion of the connection transfer at the second user equipment and the change in network utility. The operation may also include transmitting the change in network utility to the reinforcement learning model. In some implementations, the operation may further include transferring the offload instruction to the scheduler for initiating the connection transfer at the first user equipment based on the verification of the offload instruction by a centralized unit.

[0011] Depending on the implementation, the operation may include obtaining the average data utilization from the cell before selecting the first cell, where selection is based on the average data utilization and a defined policy. In the example, the defined policy is based on a function of cell load, physical resource block (PRB) utilization, and the load status of neighboring cells.

[0012] According to some implementations, activating the admission control process may include recommending an admission policy for the second user equipment to the control unit (CU). Based on the CU's acceptance of the policy, and based on the determination that the second user equipment should be admitted in the same radio unit (RU) from which the connection request was received, an acceptance confirmation may be sent to the second user equipment. Furthermore, activating the admission control process may include configuring the second user equipment using a selected cell within the cell.

[0013] In some implementations, activating the admission control process may include recommending an admission policy for the second user equipment (User Equipment) to the control unit (CU). Based on the CU's acceptance of the policy and based on the determination that the second user equipment should be admitted in a different cell than the cell from which the connection request was received, redirection information is sent to the second user equipment. Furthermore, activating the admission control process may include configuring the second user equipment using different cells. According to some implementations, the operation may include transmitting cell-level data to a reinforcement learning model based on the completion of admission for the second user equipment in different cells.

[0014] Another embodiment relates to a non-transitory machine-readable medium including executable instructions that, when executed by a processor of a network device, facilitate the execution of operations. These operations may include energy efficiency-aware load balancing performed by a system including a processor with respect to served user equipment. Energy efficiency-aware load balancing distributes served user equipment across multiple cells of a communication network. The operations may also include system-controlled admission of other user equipment to the communication network. Energy efficiency-aware load balancing and admission control may include evaluating near-real-time quality of service (QoS) performance metrics feedback. Based on the near-real-time QoS performance metrics feedback, the operations may include controlling energy efficiency-aware load balancing and admission to mitigate QoS degradation for served user equipment.

[0015] In one implementation, energy efficiency-aware load balancing may include selecting a first cell from a plurality of cells for offloading a first user equipment (UE) from its service to a second cell from the plurality of cells. The plurality of cells are under the control of a near real-time radio access network intelligent controller. Based on the selection of the first cell, energy efficiency-aware load balancing may further include providing a centralized unit with information indicating an offloading instruction for the first UE for verification.

[0016] In some implementations, admission control for other user equipment (UEs) may include activating an admission control process based on receiving a connection request from a first UE among the other UEs. Based on the outcome of the admission control process, and according to the acceptance of admission policies and utility functions, the first UE is selectively admitted to multiple cells.

[0017] To achieve the foregoing and related objectives, the disclosed subject matter includes one or more of the features described more fully below. Certain illustrative aspects of this subject matter are set forth in detail in the following description and accompanying drawings. However, these aspects indicate only a few of the various ways in which the principles of this subject matter can be employed. Other aspects, advantages, and novel features of the disclosed subject matter will become apparent from the following detailed description when considered in conjunction with the accompanying drawings. It should also be understood that the detailed description may include additional or alternative embodiments beyond those described in the summary of this invention. Attached Figure Description

[0018] Various non-limiting embodiments are further described with reference to the accompanying drawings, in which:

[0019] Figure 1 The illustration shows an example non-limiting system architecture for multi-cell user equipment admission control according to one or more embodiments described herein;

[0020] Figure 2 A first equation (1) for optimizing an objective is illustrated according to one or more embodiments described herein.

[0021] Figure 3 The second equation (2) for the power consumption factor is illustrated according to one or more embodiments described herein;

[0022] Figure 4 The illustration shows a third-party program (3) for mathematically expressing different urgency or priority for network services according to one or more embodiments described herein;

[0023] Figure 5 An example non-limiting table illustrating RRC rejection rate variations based on parameter changes according to one or more embodiments described herein is shown.

[0024] Figure 6 The illustration shows a flowchart of an example non-limiting computer implementation of a method for facilitating load balancing for multi-cell user admission control according to one or more embodiments described herein;

[0025] Figure 7 The illustration shows a flowchart of an example non-limiting computer implementation of a method for facilitating multi-cell user admission control according to one or more embodiments described herein;

[0026] Figure 8A The fourth equation (4) for the long-run reward function is illustrated according to one or more embodiments described herein.

[0027] Figure 8B The fifth equation (5) for the long-run reward function is illustrated according to one or more embodiments described herein;

[0028] Figure 9 The illustration shows a flowchart of an example non-limiting computer implementation of a method for facilitating dynamic load balancing in an advanced communication network according to one or more embodiments described herein;

[0029] Figure 10 The illustration shows a flowchart of an example non-limiting computer implementation of a method for facilitating energy-aware admission control in an advanced communication network according to one or more embodiments described herein;

[0030] Figure 11 The illustrations depict exemplary, non-limiting computing environments that may facilitate one or more embodiments described herein; and

[0031] Figure 12 The illustration depicts an example, non-limiting networking environment that may facilitate one or more embodiments described herein. Detailed Implementation

[0032] One or more embodiments are now described more fully below with reference to the accompanying drawings, in which exemplary embodiments are illustrated. In the following description, numerous specific details are set forth for purposes of explanation in order to provide a thorough understanding of the various embodiments. However, various embodiments may be practiced without these specific details. In other instances, well-known structures and devices are shown in block diagram form to facilitate the description of the various embodiments.

[0033] The high energy consumption of mobile networks (e.g., 5G and other advanced networks) is a concern for various reasons. For example, high energy consumption can increase operators' operating expenses (OPEX). In another example, high energy consumption can increase atmospheric emissions, which may directly conflict with strategic climate goals and / or policies adopted by governments and companies worldwide. Conventional static energy-saving techniques are ineffective in mobile networks with varying traffic loads and user equipment mobility patterns. Several energy-saving (ES) features, such as deep sleep modes, carrier shutdown, and radio frequency (RF) channel shutdown, can be used in conventional cellular networks (e.g., 5G and other advanced networks). However, due to the large parameter space involved in determining energy consumption, the resulting optimization problem becomes non-polynomial difficult (NP-hard), requiring extensive computation to derive optimal values ​​for these parameters.

[0034] Recently, both academia and industry standards have proposed shorter timescales for Energy Efficiency (ES). These recommendations include symbol-level, subframe-level, and / or frame-level Advanced Sleep Modes (ASM). The challenge for network operators and standardization bodies is to simplify network operation procedures for specific energy efficiency (EE) use cases, such as activating and / or deactivating sleep mode functionality and site energy management.

[0035] To overcome the aforementioned and related problems, this paper discusses a data-driven approach that outperforms classical optimization techniques in terms of performance and real-time inference. Techniques for leveraging artificial intelligence (AI) and / or machine learning (ML) for EE are provided, where the impact on quality of user experience (QoE) is negligible—a previously unexplored area of ​​knowledge related to communication networks.

[0036] In this regard, to avoid any doubt, any embodiments described herein are not limited in the context of performance optimization, and should also be considered to include any techniques for implementing essential aspects or portions of the described aspects to improve or increase performance, even if it results in suboptimal variations obtained by relaxing aspects or portions of a given implementation or embodiment.

[0037] A concern with future networks (5G, 6G, NR, etc.) is accommodating a larger number of user equipments while meeting the diverse Quality of Service (QoS) requirements of user devices (UEs). To this end and for related purposes, this paper presents a multi-cell framework for dynamic load balancing across cell clusters, minimizing network energy consumption while ensuring high QoS for UEs. The disclosed embodiments utilize reinforcement learning and other AI techniques (e.g., transfer learning, federated learning, and / or intent-based learning).

[0038] While 5G networks inherently offer improved efficiency compared to previous generations of cellular networks, their power consumption (and that of other advanced networks) is higher than that of other networks, such as Long Term Evolution (LTE) networks. This higher power consumption in 5G networks is due to the exponential growth in traffic flowing through the network and the need to process complex tasks more quickly to facilitate high target data rates. Therefore, achieving network sustainability through increased energy savings is a crucial design requirement for current and future networks.

[0039] Without any control over the number of UEs admitted to a cell, the QoS experienced at the UE may deteriorate because the number of UEs served by the cell may saturate the cell's capacity. However, due to the increasing size and diversity of devices (e.g., UEs), simply setting a fixed threshold to limit the maximum number of devices served by a cell is inefficient. Several measures have been considered in network operations. One measure is a process and criteria for setting the maximum number of UEs that can simultaneously use a network slice (NS) as part of NS admission control (AC). Another measure is that operators can set counters at the Radio Resource Control (RRC) level to control the maximum number of UEs in a cell.

[0040] Recently, some data-driven schemes based on dynamic thresholds have emerged. However, their decisions are applied to each individual cell and may lead to negative service for rejected UEs and / or network load imbalance. Accordingly, the disclosed embodiments provide a scheme for network-wide user admission control that aims to maximize the number of network-wide served users in an energy-efficient manner while satisfying user QoS constraints.

[0041] Figure 1An example non-limiting system architecture 100 for multi-cell user equipment admission control according to one or more embodiments described herein is illustrated. As mentioned, the disclosed embodiments, including admission control, can be implemented within various types of decomposed architectures.

[0042] Note that the O-RAN framework will be discussed for illustrative purposes. However, the disclosed embodiments are not limited to O-RAN framework implementations, and alternatively, other types of decomposed architectures can be utilized in conjunction with the various embodiments discussed herein. Furthermore, due to the involvement of the O-RAN framework, network devices may include, but are not limited to, O-RAN radio units (O-RUs) and random access network intelligent controllers (RICs). Additionally, network automation tools include, but are not limited to, rApp and xApp.

[0043] The disclosed embodiments provide technical solutions to several technical problems. For example, the disclosed embodiments can simultaneously achieve load balancing and energy efficiency. This document provides methods and other embodiments for load balancing of cell clusters within a network, with the goal of maximizing a utility function that, for a given number of UEs, maximizes the network's energy efficiency while ensuring that QoS constraints for different categories of UEs are met.

[0044] In another example, the disclosed embodiments provide a near real-time online learning-based approach. Conventional load balancing methods fail to gain insights from the network's current traffic conditions, nor can they combine them with predicted traffic load to optimally budget capacity within the network. These methods typically tend to rely on predefined thresholds that are used as triggers to initiate user migrations when capacity hotspots are created. Even when such triggers attempt to re-optimize network conditions, comprehensive energy consumption criteria are rarely considered. Therefore, the disclosed embodiments provide a method that can be applied to provide near real-time strategies for load balancing in an energy-efficient manner within an optimization framework that holistically considers both load balancing and energy consumption.

[0045] In yet another example, the disclosed embodiments overcome the problems associated with the lack of standardized implementations in decomposed architectures. As more decomposed network architectures (such as network devices based on Open Radio Access Networks (O-RAN)) are developed for AI and / or ML-based network optimization, appropriate message and control flows between various entities also need to be established. Therefore, from an implementation perspective, it is also necessary to explicitly define the data and control flows between the elements of an O-RAN node. Data used for the purposes of the disclosed embodiments may include statistics and network key performance indicators (KPIs) from O-RAN Radio Units (O-RUs) and O-RAN Data Units (O-DUs) to the RAN Intelligent Controller (RIC), model training, model deployment, and messages to be exchanged for user admission control and traffic steering configuration from applications deployed on the RIC platform.

[0046] As discussed herein, a data-driven approach is provided that improves network UE admission performance while dynamically balancing network load and reducing its energy consumption. This subject matter discloses an energy-efficient method for using AI and / or ML techniques within a framework to minimize QoS violations in both UE admission and load balancing (LB). The disclosed embodiments are based on two running (e.g., executing) applications: one for admission control and another for dynamic load balancing within the network. These two running applications can be executed simultaneously, concurrently, substantially simultaneously, concurrently, continuously, or at different times. The primary objective of both applications is to maximize the number of UEs requesting service within a given set of QoS constraints defined for each service class. For clusters of radio unit (RU) cells managed by a common central coordinating and / or control entity, such as a RAN intelligent controller, optimization is performed over a longer time frame (e.g., 10 time slots) and across the entire cell space than real-time algorithms (e.g., L2 scheduling).

[0047] Figure 2 A first equation (1) for optimizing an objective is illustrated according to one or more embodiments described herein. In the first equation (1), 𝜌 t It is the RRC rejection rate, and 𝜋 t It is the power consumption factor affected by a combination of the following four constraints: Constraint 1: β > α ≤ 1, Constraint 2: SINR c > SINR thresh,c , Constraint 3: Delay c < Delaythresh,c , Constraint 4: PRButil cell < PRButil thresh .

[0048] RRC rejection rate (I) t The percentage of UEs that were not admitted to the network after receiving a connection request from a UE is also included. Additionally, the power consumption factor (F) is... t The power consumed by the network during a time instance is the ratio of the maximum power consumption under peak load without any power-saving measures, and is determined by... Figure 3 The second equation (2) is given.

[0049] In the second equation (2), P Cluster yes T Average power consumption per time slot. P max This refers to the maximum power consumption of the cluster when the cell is operating at full load. Similarly, UE migration rate... σ t This represents the percentage of UEs that must be switched to a neighboring cell as part of the optimization process. C It is a group of device categories in the network. Device categories can be associated with different 5G use cases, such as enhanced mobile broadband (eMBB) devices for high data volume and throughput requirements, ultra-reliable low-latency communication (URLLC) devices for accessing latency- or delay-sensitive devices, and massive machine-type communication (mMTC) for Internet of Things (IoT) and / or Internet of Everything (IoE) based device types.

[0050] also, CELL and U These are collections containing all cells and UEs within the cluster. Furthermore, α It is a trade-off parameter between RRC rejection rate and network power consumption, and β It is a penalty-weighted constant within the utility function, used to account for the control and signaling overhead caused by UE handover within the load balancing mechanism. To avoid frequent and unnecessary handovers, β It can be retained in comparison α A higher value.

[0051] The application goal is to be in T Maximize the number of UEs served within each time slot interval, while ensuring the serviceability of this category of UEs. TThe average (or percentile) UE signal-to-interference-plus-noise ratio (SINR) over each time slot remains above a set threshold. Service latency, which is critical for UEs based on Ultra-Reliable Low-Latency Communication (URLLC), can also be used as an optimization constraint. PRButil cell This represents the Physical Resource Block (PRB) utilization of cells from the set CELL containing all cell sites in the network. This constraint states that the PRB utilization of all cells at all time slots should not exceed the value specified by the constraint. PRButil thresh The maximum PRB utilization limit is given. This strategy aims to recommend actions to achieve these goals in an energy-efficient manner by explicitly incorporating energy efficiency into the utility function. Depending on the value of , the recommended strategy establishes a Pareto optimal trade-off between RRC rejection and the power consumption of the cell cluster.

[0052] Different categories of UE devices may have different urgency or priority for network services. For example, UE categories with real-time requirements will have a more urgent need for resources compared to other categories (or UEs) with non-real-time constraint latency. This is explained in the RRC rejection rate. Figure 2 The first equation (1) is captured, where rejecting a UE with a high priority category can generate a higher penalty than rejecting a UE with a non-priority category. Mathematically, this can be given as a third process (3), such as Figure 4 As shown in the diagram.

[0053] In the third-party process (3), γC is the category " C The priority weight of "and Rt , C It is a time slot t The categories in " C "UE rejection rate. As an example, URLLC devices with strict service requirements can have a higher priority factor compared to mMTC devices. As an example, " Figure 5 An example non-limiting table 500 illustrates an RRC rejection rate variation based on parameter changes according to one or more embodiments described herein. Table 500 uses a combination of scenarios to illustrate the sensitivity of RRC rejection rate values ​​to categories with higher priority weights.

[0054] For example, Table 500 shows different categories of devices, such as URLLC devices, mMTC devices, etc. Each scenario (Scenario 1 through Scenario 6) is provided to show how it can affect the RRC rejection rate (R). t Examples of ).

[0055] The disclosed embodiments provide automatic dynamic load balancing and admission control through a novel utility function. Current industry standards already include elements such as load balancing, power control, and user admission control, which typically use threshold-based criteria. However, these thresholds are inherently static and therefore cannot adapt to real-time traffic changes and QoS violations. The disclosed embodiments provide a data-driven framework with dynamic policies for joint power consumption, user admission control, and automatic load balancing, while ensuring user QoS satisfaction. To this end, the defined utility function is a combination of total cluster power consumption, RRC rejection rate, and handover penalties triggered by load balancing.

[0056] Furthermore, the disclosed embodiments provide a data-driven framework for joint optimal admission control for energy efficiency and load balancing. In the data-driven approach provided herein, performance optimization is performed on cell clusters within the network. Due to the formulation leading to a cell-level multi-objective optimization problem, complex relationships exist between parameter sets, which, when processed independently, may result in a set of conflicting actions, further exacerbating this complexity. Due to the need for joint optimization, a mechanism is defined so that the central controller can propose suggestions from multiple strategies involving UE load balancing across cells and RRC request forwarding to improve PRB utilization, network energy efficiency, while maintaining overall user UE QoS.

[0057] Because the environment in which optimization is being performed can be highly dynamic, a reinforcement learning (RL)-based model is used, which receives network telemetry data from RU and DU nodes. This model provides user AC recommendations upon receiving a new UE admission request and load balancing suggestions at pre-specified time intervals (or via network event-based triggers). Its actions are updated based on rewards received from the environment. The RL agent perceives and interprets the environment, takes actions, and learns through trial and error to achieve optimal results according to the optimization objectives. The optimization function for each cell cluster is a combination of RRC rejection rate, cluster-wide cell power consumption, and handover cost; it is constrained by QoS satisfaction rates for each device category for the UE. Note that QoS criteria are device category independent, meaning each device category will have a unique set of KPIs and / or thresholds to determine the percentage of devices with satisfactory QoS performance. Different cells within the cluster should cooperate to maximize the cluster's cumulative rewards.

[0058] Refer again Figure 1 RAN 102 and near-RT-RIC 104 are provided. As mentioned, although O-RAN has been discussed as a deployment architecture for the purposes of describing the disclosed embodiments, similar mechanisms will also apply to other versions of decomposed network architectures.

[0059] In the O-RAN-based decomposed architecture, Figure 1 The network model in the diagram assumes that one or more RUs 106 are connected to a single RAN Data Unit (DU 108) and RAN Control Unit (CU), as illustrated in the CU Control Plane (CU-CP 110) and CU User Plane (CU-UP 112). DU 108 and CU are connected to a regional cloud including the near-RT-RIC 104. Although not illustrated, a scheduler may be included in DU 108.

[0060] Although Figure 1 The architecture depicts a one-to-one relationship between CUs (e.g., CU-CP 110 and CU-UP 112) and near-real-time RICs (near-RT-RIC 104). However, it's important to note that this framework is also applicable when deploying a single near-RT RIC to optimize network operation across multiple CUs, allowing multiple cells to be managed by xApps (software applications within the near-RT-RIC used to implement specific functions or services in near real-time) hosted by the near-RT-RIC. Near-RT-RIC 104 supports Admission Control xApp 114, which is responsible for cooperative UE admission and QoS control across multiple cells within the near-RT-RIC's coverage area. Near-RT-RIC 104 also supports Load Balancing xApp 116, which handles load balancing for UE admission. Furthermore, Near-RT-RIC 104 can improve the performance of multiple cells within its coverage area.

[0061] xApps (e.g., admission control xApp 114, load balancing xApp 116) are based on RL methods, such as the Deep Deterministic Policy Gradient (DDPG) algorithm, a model-free, online, non-policy reinforcement learning approach. DDPG agents are actor-critic (AC) RL agents that search for the optimal policy that maximizes the expected cumulative long-term reward. The actor-critic (AC) agent can implement an actor-critic process. For example, the actor-critic process can include model-free, policy-based reinforcement learning methods. The actor-critic agent can directly optimize the policy (the actor) and can use the critic to estimate the expected discounted cumulative long-term reward. DDPG agents utilize the history of past actions and rewards. In the various embodiments provided herein, action space pruning techniques can be applied to reduce the action space in both AC xApps and LB xApps.

[0062] In some embodiments, model training can be performed offline and / or on a digital twin to avoid functional network interruptions. Once the models for LB and AC use cases are trained, the inference model for cell-level collaboration is deployed as an xApp in a near-RT-RIC. RL model updates can continue based on new data to capture changes in the operational environment. However, to save the computational cost of continuous training, other mechanisms, such as model performance thresholds, can be used to trigger model retraining. For example, retraining can be performed if performance is determined to fall below a set threshold. Both xApps aim to... Figure 2 The constrained optimization problem is outlined in the first equation (1). A description of xApp will now be provided.

[0063] It should be noted that terms such as “real-time,” “near real-time,” “dynamically,” “instantaneously,” and “continuously” can refer to data collected and processed sequentially within a given context, without perceptible delay. The timeliness of the data or information is delayed only by the time required for electronic communication, the actual or near-actual time during the occurrence of a process or event, and the temporary conditions measured by real-time software, real-time systems, and / or high-performance computing systems. Real-time software and / or performance can be employed via synchronous or asynchronous programming languages, real-time operating systems, and real-time networks, each of which provides a framework for building real-time software applications. A real-time system can be a system whose application can be considered (in the context) of primary priority. In real-time processes, samples of analysis (input) and generation (output) can be continuously processed (or generated) at the same time (or nearly the same time) required for the same set of samples to be input and output, without being affected by any processing delays.

[0064] Figure 6 A flowchart illustrating an example non-limiting computer implementation of method 600 for facilitating load balancing for multi-cell user admission control according to one or more embodiments described herein is shown. The computer implementation of method 600 and / or other methods discussed herein can be implemented by a system including a processor and memory. In the example, the system can be implemented by a network device with a decomposed network architecture. Note that... Figure 6 The embodiments discussed are related to deployment within the O-RAN framework; however, the disclosed embodiments are not limited to the O-RAN framework.

[0065] Load balancing xApps according to one or more embodiments can be used to facilitate Figure 6Load balancing is implemented. At 602, the load balancing (LB) xApp is triggered (or activated). This application can be triggered after a set time interval and / or when one of the cells in the cluster reaches a utilization threshold. At 604, cell-level data is collected from the CU. The average data utilization from the cell can be determined. Based on the average data utilization from cells within the near RT-RIC control area, the xApp decides at 606 whether a UE from a particular cell needs to be offloaded to another candidate cell. At any decision instance, the xApp will select at most a single cell for UE offloading to avoid drastic changes in network load. The dynamic load balancing policy is a function of cell load, PRB utilization, and the load status of neighboring cells.

[0066] When or after selecting a cell for load balancing at point 606, the CU receives a policy notification. At point 608, it is determined whether the CU accepts the policy. If the policy is not accepted (“No”), the decision is transmitted to xApp at point 610.

[0067] Alternatively, if the CU accepts the policy (“Yes”) at 608, the CU can verify the instruction and transfer it to the MAC scheduler for initiating a connection transfer from the source cell to the target cell. For example, at 612, the CU notifies (or instructs) the CU to initiate a handover of the UE to the appropriate cell. The number of UEs that should be offloaded depends on the policies of the radio resource and mobility management layer (also known as Layer 3 (L3)) and can be a function of the UE QoS, UE Reference Signal Received Power (RSRP), and UE SINR value.

[0068] When or after UE migration is complete, the results (or rewards) (as a function of the change in network utility caused by the action) are fed back to xApp to refine and improve future policy recommendations. For example, at 614, the CU sends new cell-level data when or after load balancing is complete for action evaluation and model updates.

[0069] Figure 7 A flowchart illustrating an example non-limiting computer implementation of a method 700 for facilitating multi-cell user admission control according to one or more embodiments described herein is shown. The computer implementation of method 700 and / or other methods discussed herein can be implemented by a system including a processor and memory. In the example, the system can be implemented by a network device with a decomposed network architecture. Note that... Figure 7 The embodiments discussed are related to deployment within the O-RAN framework; however, the disclosed embodiments are not limited to the O-RAN framework.

[0070] According to one or more embodiments, Figure 7Admission control can be facilitated via an Admission Control (AC) xApp. The Admission Control xApp is an application in the model (besides the Load Balancing xApp). The Admission Control xApp acts as a gatekeeper for the number of UEs within the network.

[0071] At 702, the xApp is triggered (or activated), which can be based on receiving an RRC request at the RU. At 704, when or after the RU receives the RRC request, the RU forwards the RRC request via the CU. If it is determined that the cell cluster near the RT-RIC is already congested, the AC xApp can reject RRC requests from the UE. However, with consecutive rejections, the RRC rejection rate increases, which reduces cluster utility. When the O-RU receives an RRC request, the application (e.g., admission control xApp) is triggered.

[0072] Instead of RRC directly deciding which cell a UE should be admitted to, RRC utilizes feedback from xApp, which uses cell-level data such as cell load, power consumption, UE device class distribution, QoS values ​​for each device class, and neighboring cell load to determine which cell the UE should connect to (in the case of network admission). Accordingly, at 706, xApp recommends a policy regarding which cell the UE should be admitted to.

[0073] At 708, it is determined whether the CU accepts the policy. If it does not accept (“No”), the CU transmits the rejection decision to the xApp at 710. For example, if the xApp determines that admitting the UE to any available cell would reduce utility, it sends an RRC rejection recommendation to the O-CU.

[0074] Alternatively, if the policy is accepted at 708 (“Yes”), then at 712 it is determined whether to recommend admitting the UE in the same RU that received the RRC request. If the policy recommends admitting the UE in the cell of the same O-RU that received the RRC request (e.g., the determination at 712 is “Yes”), at 714, the O-CU sends the RRC acceptance back to the UE. Thereafter, at 716, RRC connection setup is performed. For example, the RU completes the UE's RRC connection setup using the selected cell.

[0075] Alternatively, if at 712 the determination is that xApp recommends admitting the UE to cells with different O-RUs ("No"), then at 718, RRCRedirectInfo It is sent to the UE. For example, RRC redirection information may be sent to the UE when or after the CU approves the policy. At 720, the corresponding RU completes the RRC connection setup process using the new cell identified (or selected) by xApp. At 722, when or after UE admission is completed, the CU sends new cell-level data for action evaluation and model updates.

[0076] According to some embodiments, this paper provides a RL-based learning mechanism for jointly operating AC and LB applications (e.g., admission control xApp and load balancing xApp). This embodiment includes an RL-based process, which will be discussed in further detail below. The novelty of this embodiment lies in how to use elements of the state space to define actions associated with the AC and LB applications, and an exponentially shaped reward function to ensure faster convergence. This process can be any RL-based process, such as a Deep Q-Network (DQN), or any other variant (such as DDPG), as detailed below.

[0077] State Space: The state space used for RL models can contain various features from the cell cluster, some of which are described below as a non-exhaustive list for each cell. • Cell load (number of connected UEs) •PRB utilization rate • RSRP xth percentile • SINR xth percentile • Delayed xth percentile •RRC rejection rate •N1 cell load •N1 RSRP xth percentile •N1 SINR xth percentile • N1 Delayed xth percentile •N2 cell load • N2 RSRP xth percentile • N2 SINR xth percentile • N2 delay x percentile

[0078] Near-RT-RIC will receive cell load, RSRP, SINR, and latency values ​​for the two nearest neighbor cells (N1 - nearest and N2 - second nearest) over a longer timescale. Although neighbor cell statistics may not be updated in real time, past values ​​are still expected to aid in learning, as the agent will make decisions based on comparisons of its own load and KPIs with neighbor cells. Finally, instead of using the commonly used mean statistics, the disclosed embodiments use the x-th percentile value, where "x" determines the target satisfaction level. For example, the 5th or 10th percentile statistics would mean that the embodiments ensure at least 95% or 90% of users experience satisfactory performance in terms of QoS.

[0079] Action Space: The action space for AC applications (e.g., access control xApp) is accept (ACCEPT) and the cell ID of the cell to which the application forwards the RRC request (in the case of a cell that is not receiving the RRC request); or reject (REJECT) if the application believes that adding a UE to any cell within the cluster would impair the utility function. This takes into account the existence of [missing information - likely related to cluster behavior]. N For each cell, the action space for this application (e.g., the AC application) is: N + 1 For LB applications (e.g., load balancing xApp), the action space at trigger time is to choose whether to offload some UEs to improve overall network efficiency, or to select cells without UE handover. Similar to AC applications, this applies to control clusters... N The action space size of the LB application in each cell is N + 1 .

[0080] Rewards: A unified reward function for the application (e.g., admission control xApp, load balancing xApp) reflects the utility of the objective function, supplemented by a reward shaping function to accelerate algorithm convergence. To promote even faster convergence, an exponentially based reward shaping function is applied, which generates higher rewards for actions that provide near-optimal utility values. This amplifies the differences between utility function values. Distinguishing between application actions allows for acceleration of the stochastic gradient descent (SGD) algorithm within the RL algorithm.

[0081] Figure 8A and Figure 8B The long-term reward functions are illustrated separately (e.g., equation 4 and equation 5). Figure 8B The fifth equation (5) gives the utility function u t In this context, the penalty for user migration rate imposes a higher weighted cost to avoid frequent handovers, as they introduce additional signaling loads that add to the network cost. This depends on... β The value of is only recommended when the benefits of dynamic UE handover in terms of QoS improvement and energy efficiency outweigh the signaling costs; LB applications will recommend handover strategies. In most cases, a more balanced network not only improves QoS performance but also improves overall load-related power consumption in terms of PRB utilization. Similarly, AC applications tend to achieve an optimal trade-off between network power consumption and RRC rejection rate, depending on the trade-off parameter 𝛼 managed by the network operator. When QoS degradation is too severe, the cluster can reject admission requests to improve the service quality of existing UEs. In this use case, an invalid action could be when the application recommends moving the UE to a fully loaded cell or a cell that does not provide coverage to the UE at its current location.

[0082] As discussed, this paper provides a method (and other embodiments) for energy efficiency-aware load balancing and admission control in cellular networks using near real-time QoS KPI feedback. This load balancing and admission control results in automatic load balancing and admission control that does not affect the QoS of already served UEs, rather than imposing hard-limited fixed Service Level Agreement (SLA) thresholds.

[0083] This paper also provides a method for constructing an optimized utility function to achieve the desired balance between energy efficiency and AC decisions. The utility function depends on cluster power consumption, RRC rejection rate, and the percentage of UEs migrating to neighboring cells within a known time interval. The optimized function includes UE QoS constraints on data rate and latency that depend on device type, as well as cell-level constraints on PRB utilization.

[0084] It also provides a configurable priority to control the algorithm's behavior between RRC connection rejection and cluster power consumption. Furthermore, another parameter is introduced to control the number of handovers to neighboring cells within the cluster.

[0085] Various embodiments also provide the use of data-driven algorithms within a decomposed network architecture to automate joint optimization of energy efficiency and load balancing, and to make relevant decisions on a per-cell cluster basis.

[0086] As discussed herein, systems, methods, and other embodiments for multi-cell admission control and load balancing within an O-RAN framework are provided. The goal of these various embodiments is to maximize the number of UEs admitted within a cell cluster with defined quality of service constraints, while improving energy efficiency. Several embodiments are provided to outline how these embodiments can be employed in network design. AI and / or ML technologies deployed and mapped to different network entities within the O-RAN framework are provided, hosted within the O-RAN framework along with data flows and necessary signaling for algorithm learning and policy enforcement. The disclosed embodiments provide a unique approach for simultaneously deploying AI and / or ML applications (dApp, xApp, and rApp) at the O-RAN control unit (O-CU), the near-real-time RIC radio intelligent controller (near-RT-RIC), and the non-real-time radio intelligent controller (non-RT-RIC), respectively, to enable rapid decision-making and collaboration between cells within a cell cluster. Data-driven reinforcement learning methods can be employed for online learning and real-time policy enforcement. To improve model performance, combinations of federated learning, transfer learning, and intent-based reinforcement learning methods can be leveraged to produce better results and faster convergence.

[0087] Example non-restricted non-real-time RAN intelligent controller (non-RT-RIC) functionality includes service and policy management, RAN analytics, and model training for near real-time RIC. In this regard, the non-RT-RIC implements non-real-time (e.g., first time range, such as >1 second) control of RAN elements and their resources through applications (e.g., dedicated applications called rApps). Example non-restricted near real-time RAN intelligent controller (near-RT-RIC) functionality implements near real-time optimization and control and data monitoring of O-CU and O-DU nodes within a near-RT timescale (e.g., a second time range representing a smaller time than the first time range, such as between 10 milliseconds and 1 second). In this regard, the near-RT-RIC utilizes optimization actions to control RAN elements and their resources, which typically take approximately 10 milliseconds to approximately 1 second to complete, although different time ranges can be selected. The near-RT-RIC can receive policy guidance from the non-RT-RIC and can provide policy feedback to the non-RT-RIC through a dedicated application called xApp. In this regard, the Real-Time RAN Intelligent Controller (RT RIC) is designed to process network functions on a real-time time scale (e.g., a third time range that represents a time smaller than the first and second time ranges, such as <10 milliseconds).

[0088] Figure 9 A flowchart illustrating an example non-limiting computer implementation of a method 900 for facilitating dynamic load balancing in an advanced communication network according to one or more embodiments described herein is shown. The computer implementation of method 900 and / or other methods discussed herein can be implemented by a system including a processor and memory. In the example, the system can be implemented by a network device with a decomposed network architecture. Note that... Figure 9 The embodiments discussed are related to deployment within the O-RAN framework; however, the disclosed embodiments are not limited to the O-RAN framework.

[0089] The computer-implemented method 900 can begin at 902, where the load balancing process is activated. The load balancing process can be activated at defined time instances, periodically, or at varying times. Depending on the implementation, the load balancing process can be activated based on a triggering event. For example, a triggering event could be based on the number of UEs rejected (e.g., denied service) within a defined time period exceeding an acceptable rejection threshold. In another example, a triggering event could be based on an indication of network load imbalance and / or an indication that one or more cells are more overloaded than other cells.

[0090] At 904, energy efficiency-aware load balancing is facilitated for served user equipment. Energy efficiency-aware load balancing can distribute served user equipment across a cluster of cells in the communication network. During or after the distribution of one or more served user equipment, at 906, the computer-implemented method 900 can evaluate feedback data representing near real-time service quality performance indicators. The feedback data can be near real-time service quality performance indicator feedback. Based on this feedback data, at 908, the computer-implemented method 900 can control energy efficiency-aware load balancing to mitigate service quality degradation for served user equipment.

[0091] According to some implementations, energy efficiency-aware load balancing at 904 may include selecting a first cell in the cell group for offloading a first user equipment (UE) from its service to a second cell in the cell group. This cell group is under the control of the near real-time radio access network intelligent controller. Furthermore, for this implementation, energy efficiency-aware load balancing may include providing information indicating an offloading command for the first UE to a centralized unit for verification, based on the selection of the first cell.

[0092] In addition to the implementation methods described above, based on the completion of connection transfer for user equipment, the computer-implemented method 900 may include determining the result of the connection transfer based on changes in network utility. Furthermore, changes in network utility can be transmitted for incorporation into a reinforcement learning model. According to some implementations, based on verification of the offload command by a centralized unit, the computer-implemented method 900 may transfer the offload command to the scheduler for initiating the connection transfer for the first user equipment.

[0093] In some implementations, a computer-implemented method 900 obtains the average data utilization rate from the cells before selecting the first cell. Besides this implementation, the selection of the first cell is based on the average data utilization rate and a defined strategy. For example, the defined strategy could be based on a function of cell load, PRB utilization, and the load status of neighboring cells.

[0094] Figure 10 A flowchart illustrating an example non-limiting computer implementation of a method 1000 for facilitating energy-aware admission control in advanced communication networks according to one or more embodiments described herein is shown. The computer implementation of method 1000 and / or other methods discussed herein can be implemented by a system including a processor and memory. In the example, the system can be implemented by a network device with a decomposed network architecture. Note that the embodiment of Figure 1000 is discussed with regard to deployment within an O-RAN framework; however, the disclosed embodiments are not limited to the O-RAN framework.

[0095] Note that, based on the disclosed aspects, Figure 9The computer implementation method 900 and Figure 10 The computer-implemented method 1000 can be executed at approximately the same time or at different times. Accordingly, the load balancing process and the admission control process can be in parallel (e.g., simultaneously, substantially simultaneously, concurrently) or activated (and executed) at different times. For example, one process can be activated, and at any point during its execution, another process is activated. In another example, one process can be fully executed (or almost fully executed), and then another process can be implemented (e.g., activated).

[0096] The computer-implemented method 1000 begins at 1002, where the admission control process is activated. This activation may be based on receiving one or more requests to connect to the network from one or more user devices.

[0097] At point 1004, energy-efficient control over the admission of other user equipment to the communication network is facilitated. Admission control can promote energy efficiency while maintaining a defined quality of service for existing user equipment and providing a defined quality of service for newly admitted user equipment.

[0098] When or after one or more user equipments are admitted (or denied) to the network, at 1006, the computer-implemented method 1000 can evaluate feedback data representing near real-time quality of service (QoS) performance metrics. The feedback data can be near real-time QoS performance metrics feedback. Based on this feedback data, at 1008, the computer-implemented method 1000 can control energy efficiency-aware admission control to mitigate QoS degradation for already served user equipments.

[0099] Depending on the implementation, energy efficiency-aware admission control that mitigates the degradation of service quality for already served user equipment may include controls that minimize the degradation of service quality for already served user equipment. In addition to these implementations, admission control for other user equipment may include activating an admission control process based on receiving a connection request from a first user equipment among the other user equipment. Furthermore, based on the outcome of the admission control process and based on the acceptance of admission policies and utility functions, the computer-implemented method 100 may include selectively admitting the first user equipment to the cell group. In addition to these implementations, based on the completion of admission of the first user equipment, cell-level data may be transmitted (e.g., feedback) for incorporation into a reinforcement learning model.

[0100] In some implementations, activating the admission control process may include recommending an admission policy for the second user equipment to the control unit (CU). Based on the CU's acceptance of the policy and based on the determination that the second user equipment should be admitted in the same radio unit (RU) from which the connection request was received, an acceptance confirmation may be sent to the second user equipment. Furthermore, the configuration of the second user equipment may be accomplished using a selected cell within the cell.

[0101] In some implementations, activating the admission control process may include recommending an admission policy for the second user equipment (User Equipment) to the control unit (CU). Based on the CU's acceptance of the policy and based on the determination that the second user equipment should be admitted in a different cell than the cell from which the connection request was received, redirection information is sent to the second user equipment. Furthermore, the second user equipment is configured using the different cells. These implementations may also include transmitting cell-level data to a reinforcement learning model based on the completion of the second user equipment's admission in different cells.

[0102] Referring to the flowcharts provided herein will provide a better understanding of the methods that can be implemented based on the disclosed subject matter. While these methods are shown and described as a series of processes and / or boxes for simplicity of explanation, it should be understood and recognized that the disclosed aspects are not limited by the number or order of processes and / or boxes, as some processes and / or boxes may occur substantially simultaneously with other boxes depicted and described herein in different orders. Furthermore, not all illustrated processes and / or boxes are required to implement the disclosed methods. It should be recognized that the functionality associated with the processes and / or boxes can be implemented by software, hardware, combinations thereof, or any other suitable means (e.g., devices, systems, processes, components, etc.). Additionally, it should be further recognized that the disclosed methods can be stored on an article of art to facilitate the transport and transfer of this method to various devices. Those skilled in the art will understand and recognize that the method can alternatively be represented as a series of interrelated states or events, such as in a state diagram.

[0103] The aspects of the systems, devices, apparatuses, and / or processes explained in this disclosure may constitute one or more machine-executable components embodied within one or more machines (e.g., embodied in one or more computer-readable media (or media) associated with one or more machines). When executed by one or more machines (e.g., computers, computing devices, virtual machines, etc.), one or more such components may enable one or more machines to perform the described operations.

[0104] In various embodiments, the system can be any type of component, machine, device, facility, apparatus, and / or instrument, including a processor and / or the ability to communicate effectively and / or operablely with wired and / or wireless networks. Components, machines, apparatuses, devices, facilities, and / or instruments that may include the system may include tablet computing devices, handheld devices, server-level computing machines and / or databases, laptop computers, notebook computers, desktop computers, mobile phones, smartphones, consumer appliances and / or instruments, industrial and / or commercial equipment, handheld devices, digital assistants, multimedia internet-enabled telephones, multimedia players, etc.

[0105] As used herein, the terms “storage device,” “first storage device,” “second storage device,” “storage cluster node,” “storage system,” etc. (e.g., node device) can include, for example, a private or public cloud computing system for storing data, and a system for storing data including and excluding virtual infrastructure. The term “I / O request” (or simply “I / O”) can refer to a request for reading and / or writing data.

[0106] As used herein, the term "cloud" can refer to, for example, a cluster of nodes within an object storage system (e.g., a collection of web servers) that communicate with and / or are operatively coupled to each other and host a set of applications used to serve user requests. In general, cloud computing resources can communicate with user devices via most wired and / or wireless communication networks to provide access to services based in the cloud rather than on-premises storage (e.g., on the user device). A typical cloud computing environment can include multiple layers aggregated together that interact with each other to provide resources to the end user.

[0107] Furthermore, the term "storage device" can refer to any non-volatile memory (NVM) device, including hard disk drives (HDDs), flash memory devices (e.g., NAND flash memory devices), and next-generation NVM devices, any of which can be accessed locally and / or remotely (e.g., via a storage attached network (SAN)). In some embodiments, the term "storage device" can also refer to a storage array comprising one or more storage devices. In various embodiments, the term "object" refers to a collection of user data of any size that can be stored across one or more storage devices and accessed using I / O requests.

[0108] Furthermore, a storage cluster may include one or more storage devices. For example, a storage system may include one or more clients communicating with the storage cluster via a network. The network may include various types of communication networks or combinations thereof, including but not limited to networks using protocols such as Ethernet, Internet Small Computer System Interface (iSCSI), Fibre Channel (FC), and / or wireless protocols. Clients may include user applications, application servers, data management tools, and / or testing systems.

[0109] As used herein, “entity,” “client,” “user,” and / or “application” can refer to any system or individual that can send I / O requests to the storage system. For example, an entity can be one or more computers, the Internet, one or more systems, one or more businesses, one or more computers, one or more computer programs, one or more machines, machinery, one or more actors, one or more users, one or more customers, one or more people, etc., referred to below as (one or more) entities depending on the context.

[0110] In order to provide context for various aspects of the disclosed topic, Figure 11 The following discussion is intended to provide a brief, general description of the suitable environment in which the various aspects of the disclosed topics can be implemented.

[0111] refer to Figure 11 An example environment 1110 for implementing various aspects of the above-described topics includes a computer 1112. The computer 1112 includes a processing unit 1114, system memory 1116, and a system bus 1118. The system bus 1118 couples system components, including but not limited to system memory 1116, to the processing unit 1114. The processing unit 1114 can be any available processor from a variety of available processors. Multi-core microprocessors and other multiprocessor architectures can also be used as the processing unit 1114.

[0112] The system bus 1118 can be any of several types of bus architectures, including memory bus or memory controller, peripheral bus or external bus and / or local bus using any of the various available bus architectures, including but not limited to 8-bit bus, Industry Standard Architecture (ISA), Micro Channel Architecture (MSA), Extended ISA (EISA), Intelligent Drive Electronics (IDE), VESA Local Bus (VLB), Peripheral Component Interconnect (PCI), Universal Serial Bus (USB), Advanced Graphics Port (AGP), PCMCIA, and Small Computer System Interface (SCSI).

[0113] System memory 1116 includes volatile memory 1120 and non-volatile memory 1122. The Basic Input / Output System (BIOS), containing basic routines for transferring information between components within the computer 1112 (such as during startup), is stored in the non-volatile memory 1122. By way of illustration and not limitation, the non-volatile memory 1122 may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable PROM (EEPROM), or flash memory. Volatile memory 1120 includes random access memory (RAM) used as external cache memory. By way of illustration and not limitation, RAM may be used in many forms, such as synchronous RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct Rambus RAM (DRRAM).

[0114] Computer 1112 also includes removable / non-removable, volatile / non-volatile computer storage media. Figure 11 The illustration shows, for example, a disk storage device 1124. Disk storage device 1124 includes, but is not limited to, devices such as disk drives, floppy disk drives, magnetic tape drives, Jaz drives, Zip drives, LS-100 drives, flash memory cards, or Memory Sticks. Furthermore, disk storage device 1124 may include a single storage medium or a combination of storage media with other storage media, including but not limited to optical disc drives such as compact disc ROM devices (CD-ROM), CD recordable drives (CD-R drives), CD rewritable drives (CD-RW drives), or digital versatile disk ROM drives (DVD-ROM). To facilitate connection of disk storage device 1124 to system bus 1118, a removable or non-removable interface (such as interface 1126) is typically used.

[0115] It should be recognized that, Figure 11 Software that acts as an intermediary between a user and the basic computer resources described in a suitable operating environment 1110 is described. This software includes an operating system 1128. The operating system 1128, which can be stored on a disk storage device 1124, is used to control and allocate the resources of the computer 1112. System application 1130 utilizes the resource management of the operating system 1128 through program modules 1132 and program data 1134 stored in system memory 1116 or disk storage device 1124. It should be appreciated that one or more embodiments disclosed herein can be implemented using various operating systems or combinations of operating systems.

[0116] Users input commands or information into computer 1112 through one or more input devices 1136. Input devices 1136 include, but are not limited to, pointing devices such as mice, trackballs, styluses, touchpads, keyboards, microphones, joysticks, gamepads, satellite dishes, scanners, TV tuner cards, digital cameras, digital camcorders, webcams, etc. These and other input devices are connected to processing unit 1114 via system bus 1118 through one or more interface ports 1138. Interface ports 1138 include, for example, serial ports, parallel ports, game ports, and Universal Serial Bus (USB). Output devices 1140 use some of the ports of the same type as the input devices 1136. Thus, for example, a USB port can be used to provide input to computer 1112 and output information from computer 1112 to output device 1140. Output adapters 1142 are provided to account for the existence of some output devices 1140 (such as monitors, speakers, and printers) that require special adapters, as well as other output devices 1140. By way of illustration and not limitation, output adapter 1142 includes a video card and a sound card, which provide connectivity between output device 1140 and system bus 1118. It should be noted that other devices and / or systems of devices provide both input and output capabilities, such as one or more remote computers 1144.

[0117] Computer 1112 can operate in a networked environment using a logical connection to one or more remote computers (such as one or more remote computers 1144). The one or more remote computers 1144 can be personal computers, servers, routers, network PCs, workstations, microprocessor-based appliances, peer-to-peer devices, or other common network nodes, and typically include many or all of the elements described with respect to computer 1112. For simplicity, only the memory storage device 1146 and the one or more remote computers 1144 are illustrated. The one or more remote computers 1144 are logically connected to computer 1112 via network interface 1148 and then physically connected via communication connection 1150. Network interface 1148 includes communication networks such as local area networks (LANs) and wide area networks (WANs). LAN technologies include Fiber Distributed Data Interface (FDDI), Copper Distributed Data Interface (CDDI), Ethernet / IEEE 802.3, Token Ring / IEEE 802.5, etc. WAN technologies include, but are not limited to, point-to-point links, circuit-switched networks such as Integrated Services Digital Network (ISDN) and their variants, packet-switched networks, and Digital Subscriber Line (DSL).

[0118] One or more communication connections 1150 refer to the hardware / software used to connect network interface 1148 to system bus 1118. Although communication connection 1150 is shown inside computer 1112 for clarity, it may also be outside computer 1112. For illustrative purposes only, the hardware / software required to connect to network interface 1148 includes internal and external technologies such as modems, including conventional telephone-grade modems, cable modems and DSL modems, ISDN adapters and Ethernet cards.

[0119] Figure 12 This is a schematic block diagram of a sample computing environment 1200 with which the disclosed subject matter can interact. The sample computing environment 1200 includes one or more clients 1202. Clients 1202 can be hardware and / or software (e.g., threads, processes, computing devices). The sample computing environment 1200 also includes one or more servers 1204. The server(s) 1204 can also be hardware and / or software (e.g., threads, processes, computing devices). For example, server 1204 can accommodate threads to perform transformations by employing one or more embodiments described herein. One possible communication between client 1202 and server 1204 can be in the form of data packets suitable for transmission between two or more computer processes. The sample computing environment 1200 includes a communication framework 1206, which can be used to facilitate communication between client(s) 1202 and server(s) 1204. Client(s) 1202 are operatively connected to one or more client data repositories 1208, which can be used to store information local to client(s) 1202. Similarly, one or more servers 1204 are operatively connected to one or more server data repositories 1210, which can be used to store information local to one or more servers 1204.

[0120] Throughout this specification, references to "an embodiment" or "an embodiment" mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment. Therefore, the appearance of the phrases "in one embodiment," "in one aspect," or "in an embodiment" in various places throughout the specification does not necessarily refer to the same embodiment. Furthermore, in one or more embodiments, particular features, structures, or characteristics may be combined in any suitable manner.

[0121] As used in this disclosure, in some embodiments, the terms "component," "system," "interface," "manager," etc., are intended to refer to or include computer-related entities or entities associated with operating means having one or more specific functions, wherein the entity may be hardware, a combination of hardware and software, software or software in execution, and / or firmware. As an example, a component may be, but is not limited to, a process running on a processor, a processor, an object, an executable file, a thread of execution, computer-executable instructions, a program, and / or a computer. By way of illustration and not limitation, both an application running on a server and the server itself can be components.

[0122] One or more components may reside within an executing process and / or thread, and components may be located on a single computer and / or distributed across two or more computers. Furthermore, these components may execute from various computer-readable media having various data structures stored thereon. Components may communicate via local and / or remote processes, such as according to signals having one or more data packets (e.g., data from a component that interacts with another component in a local system, a distributed system, and / or across networks such as the Internet via signals). As another example, a component may be a device having specific functions provided by mechanical parts operated by electrical or electronic circuitry, operated by a software application or firmware application executed by one or more processors, wherein the processors may be internal or external to the device and may execute at least a portion of the software or firmware application. Yet another example, a component may be a device that provides specific functions through electronic components without mechanical parts, which may include a processor for executing software or firmware that at least partially endows the electronic components with functionality. In one aspect, a component may emulate an electronic component via a virtual machine, for example within a cloud computing system. Although the various components have been described as individual components, it should be understood that, without departing from the exemplary embodiments, multiple components may be implemented as a single component, or a single component may be implemented as multiple components.

[0123] Furthermore, the words “example” and “exemplary” as used herein are meant to be used as instances or illustrations. Any embodiment or design described herein as “example” or “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments or designs. Rather, the use of the words “example” or “exemplary” is intended to present concepts in a specific manner. The term “or” as used in this application is intended to mean an inclusive “or” rather than an exclusive “or.” That is, unless otherwise specified or clearly apparent from the context, “X adopts A or B” is intended to mean any naturally inclusive arrangement. That is, if X adopts A; X adopts B; or X adopts both A and B, then in any of the above examples, “X adopts A or B” is satisfied. Furthermore, the articles “a” and “an” as used in this application and the appended claims should generally be interpreted as meaning “one or more” unless otherwise specified or clearly apparent from the context to the singular form.

[0124] Furthermore, various embodiments can be implemented using standard programming and / or engineering techniques to produce software, firmware, hardware, or any combination thereof to control a computer to implement the disclosed subject matter, methods, apparatus, or articles of art. The term "article of art" as used herein is intended to encompass computer programs accessible from any computer-readable device, machine-readable device, computer-readable carrier, computer-readable medium, machine-readable medium, computer-readable (or machine-readable) storage device / communication medium. For example, computer-readable storage media may include, but is not limited to, random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other storage technologies, solid-state drives (SSDs) or other solid-state storage technologies, magnetic storage devices such as hard disks; floppy disks; (one or more) magnetic stripes; optical discs (e.g., compact discs (CDs), digital video discs (DVDs), Blu-ray Discs™ (BD)); smart cards; flash memory devices (e.g., card, stick, key drives); and / or analog storage devices and / or virtual devices of any of the computer-readable media described above. Of course, those skilled in the art will recognize that many modifications can be made to this configuration without departing from the scope or spirit of the various embodiments.

[0125] Neither should it be assumed that the disclosed embodiments and / or aspects are exclusive to other disclosed embodiments and / or aspects, nor should it be assumed that the devices and / or structures are exclusive to the exemplary embodiments or multiple exemplary embodiments of this disclosure, unless the context explicitly states otherwise. The scope of this disclosure is generally intended to cover modifications to the depicted embodiments, including, where appropriate, the addition of content from other depicted embodiments, where appropriate, interoperability between depicted embodiments, and where appropriate, the addition of components from one(s) of embodiments to another embodiment, or the subtraction of one or more components from any depicted embodiment, where appropriate, the aggregation of elements (or embodiments) into a single device implementing aggregated functionality, or where appropriate, the distribution of functionality of a single device across multiple devices. Furthermore, combinations, combinations, or modifications of devices or elements (e.g., components) depicted herein or modified as stated above, with devices, structures, or subsets thereof not explicitly depicted herein but known in the art, or combinations, combinations, or modifications known in the art or obvious to those skilled in the art from the context of this disclosure, are also considered to be within the scope of this disclosure.

[0126] The foregoing description of the illustrative embodiments disclosed in this subject matter, including the content described in the summary of the invention, is not intended to be exhaustive of the disclosed embodiments, nor is it intended to limit the disclosed embodiments to the precise forms disclosed. While specific embodiments and examples have been described herein for illustrative purposes, various modifications are possible within the scope of such embodiments and examples, as will be appreciated by those skilled in the art.

[0127] In this regard, while the subject matter has been described herein in conjunction with various embodiments and corresponding drawings, it should be understood that other similar embodiments may be used where applicable, or modifications and additions may be made to the described embodiments to perform the same, similar, alternative, or substitute functions of the disclosed subject matter without departing from the disclosed subject matter. Therefore, the disclosed subject matter should not be limited to any single embodiment described herein, but should be interpreted in breadth and scope according to the appended claims.

Claims

1. A method comprising: A system including a processor facilitates energy efficiency-aware load balancing of served user equipment, wherein the energy efficiency-aware load balancing distributes the served user equipment across a cluster of cells in a communication network. as well as The system facilitates control over admission of other user equipment to the communication network, wherein facilitating energy efficiency-aware load balancing and facilitating the control over admission includes: The assessment represents feedback data reflecting near real-time service quality performance indicators; as well as Based on the feedback data, the energy efficiency-aware load balancing and the admission control are controlled, which mitigates the degradation of service quality for the already served user equipment.

2. The method of claim 1, wherein promoting the energy efficiency-aware load balancing comprises: Selecting a first cell in the cell group for offloading the first user equipment from the served user equipment to a second cell in the cell group, wherein the cell group is under the control of a near real-time wireless access network intelligent controller; as well as Based on the selection of the first cell, information indicating an offload command for the first user equipment is provided to the centralized unit for verification.

3. The method according to claim 2, further comprising: Based on the completion of the connection transfer of the user equipment, the system determines the result of the connection transfer according to the change in network utility; as well as The system transmits the changes in the network utility for incorporation into the reinforcement learning model.

4. The method of claim 1, wherein the control that mitigates the degradation of the service quality of the already served user equipment includes the control that minimizes the degradation of the service quality of the already served user equipment, and wherein the control that facilitates the admission of other user equipment includes: The admission control process is activated based on receiving a connection request from the first user equipment among the other user equipments. as well as Based on the results of the admission control process and the acceptance of the admission policy and utility function, the first user equipment is selectively admitted to the cell group.

5. The method according to claim 4, further comprising: Based on the completion of admission for the first user equipment, the system transmits cell-level data for incorporation into the reinforcement learning model.

6. The method of claim 1, wherein the communication network is deployed as a decomposed architecture comprising a central unit, distributed units, and a near real-time wireless access network intelligent controller.

7. The method of claim 1, wherein the cell group is configured to operate according to a new wireless network communication protocol.

8. A system comprising: processor; as well as A memory storing executable instructions that, when executed by the processor, facilitate the execution of operations, including: Activation based on the load balancing process Selecting a first cell from the cells for offloading a first user equipment to a second cell from the cells, wherein the cells are under the control of a near real-time radio access network intelligent controller; and Based on the selection of the first cell, information indicative of an offload command for the first user equipment is provided to the centralized unit for verification; and Based on receiving a connection request from the second user equipment: Activate the admission control process; and Based on the results of the admission control process, and based on the acceptance of the admission policy and utility function, the second user equipment is selectively admitted to the cell.

9. The system of claim 8, wherein the operation further comprises: Based on the completion of the connection transfer of the second user equipment, the result of the connection transfer is determined according to the change in network utility; as well as The changes in the network utility are transmitted to the reinforcement learning model.

10. The system of claim 9, wherein the operation further comprises: Based on the verification of the unloading instruction by the centralized unit, the unloading instruction is transferred to the scheduler for initiating the connection transfer of the first user equipment.

11. The system of claim 8, wherein the operation further comprises: Before selecting the first cell, the average data usage rate is obtained from the cell, wherein the selection is based on the average data usage rate and a defined strategy.

12. The system of claim 11, wherein the defined strategy is based on a function of cell load, PRB utilization, and the load status of neighboring cells.

13. The system of claim 8, wherein the load balancing process is activated based on the determination that a cell within the cluster meets a utilization threshold.

14. The system of claim 8, wherein activating the admission control process comprises: Recommend an admission policy for the second user equipment to the control unit (CU); Based on the CU's acceptance of the policy and based on the determination that the second user equipment is to be admitted in the same radio unit (RU) from which the connection request was received, an acceptance confirmation is sent to the second user equipment; as well as The second user equipment is configured using the selected cell from the aforementioned cells.

15. The system of claim 8, wherein activating the admission control process comprises: Recommend an access control strategy for the second user equipment to the control unit (CU); Based on the CU's acceptance of the policy and based on the determination that the second user equipment should be admitted in a different cell than the cell from which the connection request was received, a redirection message is sent to the second user equipment; as well as The second user equipment is configured using the different cells.

16. The system of claim 15, wherein the operation further comprises: Based on the completion of the admission by the second user equipment in the different cells, cell-level data is transmitted to the reinforcement learning model.

17. The system of claim 11, wherein the system is deployed in a decomposed architecture of a network device.

18. A non-transitory machine-readable medium comprising executable instructions that, when executed by a processor of a network device, facilitate the execution of operations, including: The system, including a processor, performs energy efficiency-aware load balancing with respect to the served user equipment, wherein the energy efficiency-aware load balancing distributes the served user equipment across multiple cells of the communication network; as well as The system controls the admission of other user equipment to the communication network, wherein the energy efficiency-aware load balancing and the control of the admission include: Evaluate feedback on near real-time service quality performance indicators; as well as Based on the near real-time service quality performance index feedback, the energy efficiency-aware load balancing and the admission process are controlled to mitigate the degradation of service quality for the already served user equipment.

19. The non-transient machine-readable medium of claim 18, wherein the energy efficiency-aware load balancing comprises: Selecting a first cell from the plurality of cells to offload the first user equipment from the served user equipment to a second cell from the plurality of cells, wherein the plurality of cells are under the control of a near real-time wireless access network intelligent controller; as well as Based on the selection of the first cell, information indicating an offload command for the first user equipment is provided to the centralized unit for verification.

20. The non-transitory machine-readable medium of claim 18, wherein controlling the access of other user equipment comprises: The admission control process is activated based on receiving a connection request from the first user equipment among the other user equipments. as well as Based on the results of the admission control process, and based on the acceptance of the admission policy and utility function, the first user equipment is selectively admitted to the plurality of cells.

Images