Network digital twin assisted network configuration generation based on performance monitoring

Abstract

Embodiments here in disclose a method and system for network configuration generation using a network digital twin (NDT) comprising a spatio-temporal graph neural network (STGNN) for performance-based simulation in a communication network system is disclosed. A simulation request is received from a management service (MnS) consumer, including network parameters related to live network nodes. Performance data is retrieved from a management server, and performance is assessed against a performance threshold. Multi-domain telecom data is processed using a spatio-temporal graph neural network (STGNN) by a feature fusion attention mechanism, trained via a hybrid feedback loop with real and synthetic data. Upon meeting performance thresholds, a configuration recommendation is generated by a configuration recommendation engine, including an explanation and confidence value. The configuration recommendation is transmitted to the MnS consumer, enabling network automation without altering the live network directly.

Description

NETWORK DIGITAL TWIN ASSISTED NETWORK CONFIGURATION GENERATION BASED ON PERFORMANCE MONITORING

[0001] The disclosure relate to a field of wireless communication. More particularly, the disclosure relates to network digital twin assisted network configuration generation based on performance monitoring in a communication network system.

[0002] Wireless communication has revolutionized the way users access information and interact by enabling connectivity without the constraints of physical cables or fixed infrastructure. The advent of wireless networks has brought about significant advancements in various fields, including mobile computing, smart home technologies, and the internet of things (IoT). These networks offer unparalleled scalability and flexibility, allowing devices to connect and communicate over vast distances without the need for physical connections.

[0003] Despite the numerous advantages, achieving optimal network performance in wireless communication remains a significant challenge. Wireless networks are vulnerable to various forms of interference, which may severely impact their efficiency and reliability. Physical obstructions such as walls, buildings, and natural landscapes may impede signal propagation, leading to weakened signal strength or complete signal loss. Electromagnetic noise from household appliances, industrial equipment, and other electronic devices may introduce additional interference, further degrading signal quality. Further, competing wireless signals from neighbouring networks or other wireless devices may cause congestion and collisions, exacerbating the problem.

[0004] As user demand fluctuates and more devices are added to the network, maintaining performance becomes increasingly complex. The dynamic nature of wireless networks means that they must adapt to changing conditions, such as varying user locations, device mobility, and fluctuating traffic loads. Traditional methods of configuring wireless networks often rely on trial-and-error approaches, which are both time-consuming and inefficient. Network administrators manually adjust settings and parameters, hoping to achieve an optimal configuration. However, this process frequently results in suboptimal performance, service degradation, network congestion, and even system failures.

[0005] Current approaches to addressing these challenges lack the sophistication required to dynamically and intelligently optimize network performance. The reliance on manual configuration and static settings fails to account for the real-time variations and complexities inherent in wireless communication. Lack spatio-temporal, topology-aware model structure, cannot fuse heterogeneous features with domain-aware attention, do not dynamically adapt weights / edges based on simulation / real feedback. As a result, users experience inconsistent connectivity, reduced data transfer rates, and increased latency, all of which hinder the overall user experience.

[0006] The above information is presented as background information only to assist with an understanding of the disclosure. No determination has been made, and no assertion is made, as to whether any of the above might be applicable as related art with regard to the disclosure.

[0007] Aspects of the disclosure are to address at least the above-mentioned problems and / or disadvantages and to provide at least the advantages described below. Accordingly, an aspect of the disclosure is to provide a method for predicting network congestion in a wireless network system.

[0008] Additional aspects will be set forth in part in the description which follows and, in part, will be apparent from the description, or may be learned by practice of the presented embodiments.

[0009] The disclosure is to provide a method and system for network digital twin (NDT) assisted network configuration generation based on performance monitoring in a communication network system.

[0010] The disclosure is to provide a NDT assisted solution for performance management in communication networks. The NDT utilizes real-time and historical network performance data along with configuration management information to simulate the network's behavior and provide automated simulation-driven network configuration recommendations.

[0011] The disclosure is to provide a modular graph ML block captures both spatial and temporal cross-node dependencies, attention fusion layer adapts to scenario / context, hybrid simulation / online training loop ensures robustness to rare / noisy events.

[0012] The disclosure is to provide an NDT system comprising a spatio-temporal graph neural network (STGNN) designed to process heterogeneous, multi-domain telecom data, network generates configuration recommmendations for network automation with integrated explanation generation.

[0013] The disclosure is to provide a graph neural network (GNN) incorporates a feature fusion attention mechanism and is trained using a hybrid loop involving real and synthetic (simulation-generated) network states.

[0014] The disclosure is to provide a method that integrates simulation and prediction to deliver network configurations without real-time network control, enhancing operational flexibility, integrates external data for comprehensive performance assessment.

[0015] In an embodiment, the disclosure provides a method for NDT assisted network configuration generation based on performance monitoring comprising receiving by an NDT server a request for simulation from a management service (MnS) consumer device associated with a live network wherein the request may comprise a plurality of network parameters associated with network nodes in the live network to be simulated, retrieving by the NDT server performance data, fault and configuration data are also collected from a management server based on the plurality of network parameters wherein the management server manages the network nodes, determining by the NDT server a performance of the network nodes based on the retrieved performance data, fault and configuration data, the network configuration recommendation from a configuration recommendation engine, wherein the performance of the network node meets a performance threshold, and transmitting by the NDT server, the network configuration recommendation to the MnS consumer.

[0016] In an embodiment, the disclosure provides a method for generating dynamic network configuration recommendations using a network digital twin (NDT) server in a communication network system, the method comprises receiving, by the NDT server, a simulation request from a management service (MnS) consumer associated with a live communication network, wherein the request comprises a plurality of network parameters associated with network nodes, retrieving, by the NDT server, network-related data including performance data from a performance assurance system, fault data from a fault management system, and configuration data from a configuration management system, each of which is part of a management server, obtaining, by the NDT server, external environmental data from an external system, generating, by the NDT server, fused feature embeddings by processing the retrieved data using domain-specific encoders and a scenario-aware gating mechanism, constructing, by the NDT server, a spatio-temporal graph of network elements using a spatio-temporal graph neural network (STGNN) inference engine, and applying message-passing operations to predict network states and generate configuration recommendations, training, by the NDT server, the STGNN inference engine using a hybrid learning loop incorporating real network data and validated synthetic data, evaluating, by the NDT server, the generated recommendations to determine feasibility, compliance, and confidence score, transmitting, by the NDT server, configuration recommendations to the MnS consumer for deployment in the live communication network.

[0017] In an embodiment, the disclosure provides a network digital twin (NDT) server for NDT assisted network configuration generation based on performance monitoring. The NDT server comprises memory storing instructions and at least one processor communicatively coupled to the memory. The instructions, when executed by the at least one processor individually or collectively, cause the NDT server to receives a request for simulation from the MnS consumer operably connected to a live network wherein the request comprises a plurality of network parameters associated with network nodes in the live network to be simulated, retrieves performance data from a management server based on the plurality of network parameters wherein the management server manages the network nodes, determines a performance of the network nodes based on the retrieved performance data, fault and configuration data, determines whether the real-time performance of the network node meets a performance threshold, receives the network configuration recommendation from a configuration recommendation engine, wherein the performance of the network node meets a performance threshold, and transmits the network configuration recommendation to the MnS consumer.

[0018] In an embodiment, the disclosure provides a network digital twin (NDT) server for generating dynamic network configuration recommendations in a communication network system. The NDT server comprises memory storing instructions and at least one processor communicatively coupled to the memory. The instructions, when executed by the at least one processor individually or collectively, cause the NDT server to receives a simulation request from a management service (MnS) consumer associated with a live communication network, wherein the request comprises a plurality of network parameters associated with network nodes, retrieves, network-related data including performance data from a performance assurance system, fault data from a fault management system, and configuration data from a configuration management system, each of which is part of a management server, obtains, external environmental data from an external system, generate fused feature embeddings by processing the retrieved data using domain-specific encoders and a scenario-aware gating mechanism, constructs, spatio-temporal graph of network elements using a spatio-temporal graph neural network (STGNN) inference engine, and apply message-passing operations to predict network states and generate configuration recommendations, trains the STGNN inference engine using a hybrid learning loop incorporating real network data and validated synthetic data, evaluates the generated recommendations to determine feasibility, compliance, and confidence score, transmits configuration recommendations to the MnS consumer for deployment in the live communication network.

[0019] Other aspects, advantages, and salient features of the disclosure will become apparent to those skilled in the art from the following detailed description, which, taken in conjunction with the annexed drawings, discloses various embodiments of the disclosure.

[0020] The above and other aspects, features, and advantages of certain embodiments of the disclosure will be more apparent from the following description taken in conjunction with the accompanying drawings, in which:

[0021] Fig. 1a illustrates the configuration of the 5G network being managed sub optimally according to the prior art.

[0022] Fig. 1b illustrates the system for managing a 5G network according to the embodiments as disclosed herein.

[0023] Fig. 1c illustrates the determination of optimum TDD Interference Mitigation Solution & Schedule by the operator according to the embodiments as disclosed herein.

[0024] Fig. 1d illustrates the block diagram of an NDT (MnS Producer), various components according to the embodiments as disclosed herein.

[0025] Fig. 2 illustrates the operator receiving the optimum TDD interference mitigation solution and schedule from a configuration generation system management according to embodiments as disclosed herein.

[0026] Fig. 3 is a block diagram indicating the components of a live communication network system in accordance with an embodiment of the disclosure.

[0027] Fig. 4 illustrates a flow chart indicating operations for NDT assisted network configuration generation based on performance monitoring recommendation in a communication network system according to embodiments as disclosed herein.

[0028] Fig. 5 illustrates a flow chart indicating operations followed by the MnS consumer in a communication network system in accordance with an embodiment of the disclosure.

[0029] Fig. 6 illustrates about the sequence diagram, a flow of sequences performed by "MnS Producer (NDT)" block to obtain network configuration recommendations to obtain network configuration recommendations according to embodiments as disclosed herein.

[0030] Fig. 7 illustrates the prediction of TDD interference in a communication network in accordance with an embodiment of the disclosure.

[0031] Fig. 8 illustrates the NDT-based predictive maintenance of hardware according to embodiments as disclosed herein.

[0032] Fig. 9 illustrates the Digital Twin-based predictive maintenance of hardware in accordance with an embodiment of the disclosure.

[0033] Throughout the drawings, it should be noted that like reference numbers are used to depict the same or similar elements, features, and structures.

[0034] The following description with reference to the accompanying drawings is provided to assist in a comprehensive understanding of various embodiments of the disclosure as defined by the claims and their equivalents. It includes various specific details to assist in that understanding but these are to be regarded as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the various embodiments described herein can be made without departing from the scope and spirit of the disclosure. In addition, descriptions of well-known functions and constructions may be omitted for clarity and conciseness.

[0035] The terms and words used in the following description and claims are not limited to the bibliographical meanings, but, are merely used by the inventor to enable a clear and consistent understanding of the disclosure. Accordingly, it should be apparent to those skilled in the art that the following description of various embodiments of the disclosure is provided for illustration purpose only and not for the purpose of limiting the disclosure as defined by the appended claims and their equivalents.

[0036] It is to be understood that the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. Thus, for example, reference to "a component surface" includes reference to one or more of such surfaces.

[0037] It should be understood at the outset that although illustrative implementations of the embodiments of the disclosure are illustrated below, the disclosure may be implemented using any number of techniques, whether currently known or in existence. The disclosure should in no way be limited to the illustrative implementations, drawings, and techniques illustrated below, including the design and implementation illustrated and described herein, but may be modified within the scope of the appended claims along with their full scope of equivalents.

[0038] The term "some" as used herein is defined as "none, or one, or more than one, or all." Accordingly, the terms "none," "one," "more than one," "more than one, but not all" or "all" would all fall under the definition of "some." The term "some embodiments" may refer to no embodiments, to one embodiment or to several embodiments or to all embodiments. Accordingly, the term "some embodiments" is defined as meaning "no embodiment, or one embodiment, or more than one embodiment, or all embodiments."

[0039] The terminology and structure employed herein is for describing, teaching, and illuminating some embodiments and their specific features and elements and does not limit, restrict, or reduce the spirit and scope of the claims or their equivalents.

[0040] More specifically, any terms used herein such as but not limited to "includes," "comprises," "has," "consists," and grammatical variants thereof do NOT specify an exact limitation or restriction and certainly do NOT exclude the possible addition of one or more features or elements, unless otherwise stated, and furthermore must NOT be taken to exclude the possible removal of one or more of the listed features and elements, unless otherwise stated with the limiting language "MUST comprise" or "NEEDS TO include."

[0041] Whether or not a certain feature or element was limited to being used only once, either way, it may still be referred to as "one or more features" or "one or more elements" or "at least one feature" or "at least one element." Furthermore, the use of the terms "one or more" or "at least one" feature or element does NOT preclude there being none of that feature or element, unless otherwise specified by limiting language such as "there NEEDS to be one or more . . ." or "one or more element is REQUIRED."

[0042] Unless otherwise defined, all terms, and especially any technical and / or scientific terms, used herein may be taken to have the same meaning as commonly understood by one having ordinary skill in the art.

[0043] As used herein, each of such phrases as "A or B," "at least one of A and B," "at least one of A or B," "A, B, or C," "at least one of A, B, and C," and "at least one of A, B, or C," may include any one of, or all possible combinations of the items enumerated together in a corresponding one of the phrases. It is to be understood that if an element (e.g., a first element) is referred to, with or without the term "operatively" or "communicatively", as "coupled with," "coupled to," "connected with," or "connected to" another element (e.g., a second element), it means that the element may be coupled with the other element directly (e.g., wired), wirelessly, or via a third element.

[0044] It should be appreciated that the blocks in each flowchart and combinations of the flowcharts may be performed by one or more computer programs which include computer-executable instructions. The entirety of the one or more computer programs may be stored in a single memory device or the one or more computer programs may be divided with different portions stored in different multiple memory devices.

[0045] Any of the functions or operations described herein can be processed by one processor or a combination of processors. The one processor or the combination of processors is circuitry performing processing and includes circuitry like an application processor (AP, e.g., a central processing unit (CPU)), a communication processor (CP, e.g., a modem), a graphical processing unit (GPU), a neural processing unit (NPU) (e.g., an artificial intelligence (AI) chip), a wireless-fidelity (Wi-Fi) chip, a BluetoothTMchip, a global positioning system (GPS) chip, a near field communication (NFC) chip, connectivity chips, a sensor controller, a touch controller, a finger-print sensor controller, a display drive integrated circuit (IC), an audio CODEC chip, a universal serial bus (USB) controller, a camera controller, an image processing IC, a microprocessor unit (MPU), a system on chip (SoC), an IC, or the like.

[0046] Referring now to the drawings, and more particularly to Figs. 1-9 where similar reference characters denote corresponding features consistently throughout the figures, there are shown preferred embodiments.

[0047] Fig. 1a illustrates a configuration of a 5G network being managed sub optimally according to the prior art. The system may include a 5G live network (102) and a management system (104). The management system (104) is responsible for overseeing the performance and configuration of the 5G live network (102).

[0048] In the prior art configuration, the management system (104) inputs configuration settings into the 5G live network (102). This process is depicted by the arrow labeled "Input configuration." The 5G live network (102) operates based on these settings, and its performance is monitored. This monitoring process is indicated by the arrow labeled "Performance Monitoring".

[0049] The performance data collected from the 5G live network (102) is fed back to the management system (104). The management system (104) analyzes this data to determine if the network is operating optimally. If the performance is found to be suboptimal, the management system (104) adjusts the configuration settings and inputs them back into the 5G live network (102). This iterative process of configuration and performance monitoring is depicted by the cyclic arrows and the label "System Configuration done repetitively for Optimum performance".

[0050] Despite the continuous monitoring and configuration adjustments, the prior art system often fails to achieve optimal performance due to the reactive nature of the management system (104). The adjustments are made after performance issues are detected, leading to periods of suboptimal operation. This reactive approach results in inefficiencies and potential disruptions in the 5G live network (102).

[0051] Fig. 1b illustrates the system for managing a 5G network, according to the embodiments as disclosed herein. The system may include a 5G live network (102), a human operator (106), and several interconnected systems forming a digital twin environment.

[0052] The 5G live network is responsible for real-time communication and data transmission. It collects performance metrics and operational data, which are used for monitoring and optimizing network performance. The data collection system interfaces with the 5G live network to gather this data, ensuring that the digital twin has up-to-date information for analysis.

[0053] The digital twin environment may include several interconnected systems, including learning, prediction, storage, simulation, virtual replicas, and modeling. These systems collaboratively work to simulate, analyze, and predict the performance of the 5G network.

[0054] The learning system processes the collected data to understand patterns and trends within the network's performance. This system utilizes machine learning algorithms to identify key performance indicators (KPIs) and their impact on network operations.

[0055] The prediction system uses the insights gained from the learning system to forecast future network performance. It anticipates potential issues and suggests proactive measures to maintain network functionality.

[0056] The storage system securely stores all collected data and analysis results. This system ensures that historical data is available for reference and further analysis, aiding in long-term network management and optimization.

[0057] The simulation system creates virtual replicas of the 5G network. These replicas allow for testing and validation of various network configurations and scenarios without impacting the live network. The simulation results provide insights into the potential impact of different configurations on network performance.

[0058] The virtual replicas system is closely linked with the simulation module. It generates detailed virtual models of the 5G network, enabling comprehensive performance monitoring and impact analysis for specific KPIs. This system helps in understanding how changes in network configuration may affect overall performance.

[0059] The modeling system integrates the insights from the learning, prediction, simulation, and virtual replicas modules. It provides the human operator (106) with optimal network configurations based on the analyzed data. The human operator (106) inputs performance KPIs and receives recommendations for network adjustments to achieve optimal performance.

[0060] The human operator (106) input performance KPIs into the modeling system and receive optimal configuration suggestions. The operator implements these configurations in the 5G live network, ensuring that the network operates efficiently.

[0061] In an embodiment, the disclosure solves when the system notices that the network is slowing down or experiencing poor signal quality, it immediately tries to predict the cause and find the best solution. for example, if a sudden spike in traffic makes the network slow, it might suggest boosting bandwidth or adjusting signal strength. Further, if bad weather is likely to disrupt signals, it may recommend changes to keep the network stable, instead of manually reacting to issues, the NDT system monitors network performance continuously, analyzes data using AI / ML, and suggests optimal configuration changes before problems significantly impact service quality.

[0062] Fig. 1c illustrates the determination of the optimum time division duplexing (TDD) Interference Mitigation Solution and Schedule by the operator (106) according to the embodiments. The 5G network (102) is intended to meet the network demands of the human operator (106), an MnS consumer (108), a configuration recommendation engine (CRE) (110), network management (112), service management (114), etc. The MnS consumer (108) represents any entity that accesses or engages with a management service delivered by a Management server (116) may include 3 main components performance assurance system (116a) (PA MnS Producer), fault management system (117) (FM MnS Producer), configuration management system (118) (FM MnS Producer), This could encompass a range of entities, including other management functions (MnFs), software applications, digital tools, or third-party organizations, provided they possess the required authorization and authentication. Further, configuration management (118) provides data related to network configuration, and weather data is received from a weather system (120). The human operator (106) utilizes a trial-and-error method to self-detect network performance bottlenecks and adjust the necessary network configuration settings to regain performance, resulting in suboptimal network performance, service degradation, congestion, and even network failures. Further, the techniques available in the prior art are not effective in managing network performance.

[0063] Fig.1d illustrates the block diagram of an NDT (MnS Producer), various components for NDT assisted network configuration generation based on performance monitoring.

[0064] Fig. 1d illustrates a block diagram of a network configuration recommendation engine (110) integrated within a management domain, according to embodiments as disclosed herein. The system is configured to generate dynamic configuration recommendations based on a variety of input sources, including performance, fault, configuration, weather, and simulation data. The network configuration recommendation engine (110) forms part of a closed-loop intelligent configuration management framework using a machine learning (ML)-based system.

[0065] The system may comprise multiple MnS (management service) producers and consumers. A management server (116) is communicatively coupled with the configuration recommendation engine (110) and consists of three primary systems―performance assurance system (116a) (PA MnS Producer), fault management system (117) (FM MnS Producer), and configuration management system (118) (CM MnS Producer). These systems collect network-related data including performance metrics, fault logs, and configuration parameters. Additional external data may be obtained from an external system (120) and a weather system, enhancing the context-awareness of the network configuration process. These inputs are fed into the engine via interfaces supporting PM / FM / CM and weather data.

[0066] The system also may comprise human operators (106), network management systems, service management systems, and self organizing network (SON) functions, which act as MnS consumers (108). These entities may request simulation of network scenarios and receive recommended configurations. The MnS consumer (108) interfaces with the network digital twin (NDT) server (125), which acts as an MnS Producer.

[0067] The NDT server (125) may include several interconnected systems. The scenario policy controller (125a) receives vector-based features and scenario data to generate a fusion or gating policy. This policy is used by the feature fusion layer (125b) to combine various feature inputs into a cohesive set of fused features. These features are fed into the ST-GNN inference engine (125c), which stands for spatial-temporal graph neural network, a model used to predict and recommend network configurations. This inference engine may also incorporate a hybrid learning loop involving simulation and real-world feedback.

[0068] The output from the ST-GNN inference engine (125c) is provided to the ML-based recommendation system (125d), which processes the predictions and generates configuration recommendations. These recommendations are evaluated by the constraints & confidence gate (125f). This system projects constraints and calculates a confidence score for each recommendation using a constraint projection confidence score.

[0069] Further, a simulation feedback trainer (125e) forms part of the hybrid learning loop, enabling continuous refinement of the ML model based on simulated and real feedback. The learning process ensures that the configuration recommendation engine (110) adapts to changing network conditions and external factors like weather.

[0070] The scenario policy controller (125a) is a critical, smart supervisory system embedded within the NDT's data processing and ST-GNN inference pipeline. Its purpose is to detect operational scenarios based on current network conditions and dynamically adapt multiple internal components of the NDT platform―especially the fusion layer and ST-GNN inference engine.

[0071] The scenario policy controller (125a) detects real-time operating conditions from multi-domain inputs. Dynamically configures fusion weighting and ST-GNN inference policies per scenario.

[0072] The scenario policy controller (125a) enables multi-policy adaptive GNN architectures with switching attention and message-passing systems.

[0073] The scenario policy controller (125a) prevents noisy data impact via gating.

[0074] The scenario policy controller (125a) supports robustness to evolving networks and new scenario integration.

[0075] The scenario policy controller (125a) continuously ingests fused signals from performance management (PM), fault management (FM), configuration management (CM), and external environmental data streams (e.g., weather, topology).

[0076] In an embodiment, the method may include the feature fusion layer (125b) applies temporal alignment using one or more of temporal convolutions or recurrent neural units to synchronize heterogeneous domain data arriving at different timescales.

[0077] In an embodiment, the method may include the ST-GNN inference engine (125c) dynamically constructs or deletes edges in the spatio-temporal graph based on real-time triggers such as KPIs, alarm signals, interference, or energy metrics.

[0078] In an embodiment, the method may include the scenario policy controller (125a) supervises the feature fusion layer (125b) and the ST-GNN inference engine (125c) by dynamically adjusting fusion weights and selecting message-passing policies based on detected network scenarios.

[0079] In an embodiment, the method may include the simulation feedback trainer (125e) filters low-quality synthetic events and prioritizes rare but critical network events to fine-tune the ST-GNN inference engine (125c) in near real-time.

[0080] In an embodiment, the method may include the constraints & confidence gate (125f) detects a constraint projection confidence score for each recommendation, and selects those that meet specified performance, risk, and policy criteria for final deployment.

[0081] In an embodiment, variables such as KPI metrics (e.g., CQI, SINR, drop rates, throughput), alarm and fault statuses, dynamic configuration parameters (e.g., handover thresholds, power settings), and exogenous factors (e.g., temperature, atmospheric pressure, humidity).

[0082] The feature fusion layer (125b) may include raw input data from each domain is first processed by a dedicated encoder system (e.g., a small multi-layer perceptron or embedding layer).

[0083] The feature fusion layer (125b) transforms raw, often heterogeneous signals into normalized feature embeddings, enabling compatibility and comparability across domains.

[0084] The feature fusion layer (125b) transforms raw, often heterogeneous signals into normalized feature embeddings, enabling compatibility and comparability across domains.

[0085] In an embodiment, the examples for heterogeneous signals into normalized feature embeddings,

[0086] For example:

[0087] PM data like KPIs, SINR, error rates encoded into a continuous embedding vector.

[0088] CM data like current handover settings or antenna parameters embedded as discrete or categorical embeddings.

[0089] Alarm and fault signals converted into binary or temporal event embeddings.

[0090] Weather and hardware data transformed via physical models and contextual embeddings.

[0091] The feature fusion layer (125b), the network continuously detects the current operational scenario (e.g., energy alert, interference storm, handover disruption).

[0092] For each domain embedding, a gating mechanism computes scenario-dependent attention weights. This gating selectively amplifies or suppresses certain domain inputs during fusion based on scenario relevance.

[0093] The gated embeddings are combined via concatenation and optionally learned linear projections and often, streams arrive at different timescales; fusion layer handles time alignment or smoothing (e.g. via temporal convolutions or recurrent units) to provide coherent temporal context to the GNN.

[0094] The ST-GNN inference engine (125c) processes fused multi-domain features across dynamically constructed graphs representing telecom network elements over time. It outputs predictive embeddings of network states and key performance indicators (KPIs), driving actionable network configuration recommendations.

[0095] The ST-GNN inference engine (125c) Continuously monitors KPI changes and triggers edge creation / deletion using configurable threshold functions. The graph adapts as network conditions evolve, enabling the model to reason over the current effective interaction topology, not just physical links.

[0096] The ST-GNN inference engine (125c) incorporates multiple specialized message passing systems trained for different scenarios (e.g., interference, handover, energy).

[0097] The ST-GNN inference engine (125c) supports on-the-fly edge construction / deletion based on live, multi-domain triggers ― a technical leap enabling focused, context-aware reasoning over actual network dynamic interactions.

[0098] In an embodiment, the system dynamically switches between multiple learned reasoning pathways (message / attention functions) depending on detected network conditions, allowing tailored inference mechanisms instead of one-size-fits-all ― greatly enhancing performance and explainability.

[0099] The simulation feedback trainer (125e) (hybrid learning loop) accelerates model learning by mixing live network data with realistic synthetic events. Includes smart filters to reject poor synthetic data, preserving model fidelity. Employs smart data prioritization to learn rare but critical telecom incidents.

[0100] The simulation feedback trainer (125e) enables the model to rapidly adapt to new network challenges, ensuring high-quality real-time recommendations. The simulation feedback trainer (125e) provides foundational support for a scenario-adaptive, multi-policy ST-GNN inference engine to maintain performance across diverse use cases.

[0101] The simulation feedback trainer (125e) Supports online fine-tuning with minimal latency to improve foresight into transient events and network anomalies.

[0102] The constraints & confidence gate (125f) translates ST-GNN embeddings and forecasts into actionable network configuration recommendations. Further ensures every recommendation is feasible and compliant, preventing risks in live telecom environments. Furthermore, dynamically adjusts confidence thresholds by operational scenario, allowing adaptive trust management sensitive to network conditions.

[0103] The constraints & confidence gate (125f) integrates multi-objective utility scores, balancing performance gains, risk reduction, operational constraints, and business priorities, to optimally select configurations. Outputs comprehensive metadata on rationale, confidence factors, and expected impact to enhance operator trust and regulatory acceptance. Tracks success / failure of implemented recommendations, feeding results back to the simulation trainer for continuous model improvement and risk minimization.

[0104] The configuration recommendation engine (110) (3GPP MDAS) serves as the finisher, that translates NDT's ML-based recommendations into network-compatible configuration commands, ensuring compliance with 3GPP standards and operator policies.

[0105] Fig. 2 illustrates the operator receiving the optimum TDD interference mitigation solution and schedule from a configuration generation system management according to embodiments as disclosed herein. The system may comprise multiple interconnected components that work together to manage network configurations dynamically.

[0106] The human operator is identified as the human operator (106). The human operator (106) interacts with the system to receive recommendations and manage network configurations. The human operator (106) interfaces with the MnS consumer (108), which is responsible for consuming the dynamic configuration recommendations provided by the system.

[0107] The MnS consumer (108) receives dynamic configuration recommendations from the dynamic configuration recommendation system (126b) of an NDT server (124)-assisted network configuration generation system management (126). The NDT server (124)-assisted network configuration generation system management (126) is responsible for generating and managing dynamic network configurations based on various inputs.

[0108] The NDT server (124)-assisted network configuration generation system management (126) may include dynamic performance management system (126a), which monitors and manages network performance dynamically. This system (126a) ensures that the network operates optimally by adjusting configurations based on real-time performance data.

[0109] The management server (116) includes performance assurance system (116a) (PA MnS Producer), fault management system (117) (FM MnS Producer), configuration management system (118) (FM MnS Producer), collects and processes performance data from the network (102). This management server (116) provides information to the dynamic performance management system (126a), enabling it to make informed decisions about network configurations.

[0110] The system configuration management system (118) is responsible for managing network configurations. It stores and updates configuration data, ensuring that the network operates according to the settings and recommendations.

[0111] The weather system (120) provides environmental data that may impact network performance. This data is used by the dynamic performance management system to adjust configurations in response to changing weather conditions, ensuring optimal network performance.

[0112] The network management system (112) oversees the overall management of the network (102). It ensures that the network (102) operates smoothly and efficiently by coordinating various management activities and implementing configuration recommendations.

[0113] The service management system (114) handles the management of network services. It ensures that services are delivered effectively and efficiently, meeting the needs of users and maintaining high service quality.

[0114] The CRE (110) is responsible for managing data and analytics related to 3GPP standards and also used as configuration recommendation. It provides insights and data that support the dynamic configuration recommendation process.

[0115] In an embodiment, the CRE (110) is hardware designed for NDT assisted network configuration generation based on performance monitoring.

[0116] In an embodiment, the CRE (100) sends optimized configurations, may include changes in antenna gain, transmission power, or interference control, to improve performance. The 3GPP MDAS is utilized as configuration recommendation engine.

[0117] The NDT server (124)-assisted network configuration generation system management (126), which includes the dynamic performance management system (126a) and dynamic configuration recommendation system (126b), works collaboratively with the management server (116), performance assurance system (116a) (PA MnS Producer), fault management system (117) (FM MnS Producer), configuration management system (118) (FM MnS Producer),, and weather system (120) to provide the human operator (106) with optimal TDD interference mitigation solutions and schedules.

[0118] Fig. 3 illustrates a block diagram indicating the components of a live communication network system according to embodiments as disclosed herein. The communication network system may comprise an NDT server (124), a MnS consumer (108), a management server (116), may include performance assurance system (116a) (PA MnS Producer), fault management system (117) (FM MnS Producer), configuration management system (118) (FM MnS Producer), an external weather system (120), and a CRE (110). All these components are connected to a live network (102).

[0119] Examples of the live network (102) include, but are not limited to, cellular networks (such as 2G, 3G, 4G, 5G, Beyond 5G (B5G) / 6G, or advanced cellular networks), local area networks (LANs) (such as Wi-Fi, Li-Fi, etc.), personal area networks (PANs) (such as Bluetooth, Zigbee, Z-Wave, etc.), wide area networks (WANs) (such as satellite communication networks, long range wide area network, narrowband IoT, low-bandwidth communication for IoT, etc.), metropolitan area networks (MANs), machine-to-machine (M2M), Ad Hoc and mesh networks, and emerging and advanced networks.

[0120] In an embodiment, the NDT server (124) may include memory (124a), a processor (124b), and a network configuration recommendation controller (124c). The memory (124a) stores information about the MnS consumer (108). The network configuration recommendation controller (124c) is connected to the memory (124a) and the processor (124b). An innovative integrated circuit, the network configuration recommendation controller (124c) may include a multi-core architecture that enables dynamic adjustment of a location reporting trigger in a communication system. Each core is optimized for specific tasks for NDT assisted network configuration generation based on performance monitoring recommendation in a communication network system.

[0121] A request for simulation is received by the network configuration recommendation controller (124c) from the MnS consumer (108) operably connected to the live network (102). The MnS consumer (108) is responsible for managing and supervising the network services and ensuring optimal performance. Detailed specifications about the network nodes, such as their current configuration, operational status, and historical performance data, are included in the request for simulation. Performance data is retrieved from a management server (116) based on a plurality of network parameters to determine the real-time performance of the network nodes. The Management server (116) monitors various metrics such as bandwidth utilization, error rates, and latency to provide accurate and up-to-date performance data. The network configuration recommendation controller (124c) further determines whether the real-time performance of the network node meets a performance threshold by comparing the current performance metrics against specified thresholds to identify any deviations or areas for improvement. When the performance of the network node meets a performance threshold, the network configuration recommendation is received from the integrated CRE (110) and transmitted to the MnS consumer (108). Advanced algorithms and machine learning techniques are utilized by the CRE (110) to generate optimal configuration recommendations based on the analyzed data.

[0122] The network configuration recommendation controller (124c) receives a request for simulation from the MnS consumer (108). The request may comprise a plurality of network parameters associated with network nodes in the live network (102) to be simulated. These parameters include detailed information about the network topology, node capabilities, and current traffic patterns. The plurality of network parameters associated with network nodes may comprise at least one of a simulation scope defining target network nodes to be simulated, a performance management threshold defining target thresholds for performance metrics (PMs) to be achieved by the network configuration recommendation, a target quality of service (QoS) defining quality of service parameters for the network configuration recommendation, a granularity defining a granularity period for collecting the performance data, a target load defining a target load of the network nodes to be simulated, a target time defining a time at which the performance of the network nodes needs to be collected, and a target energy-saving parameter indicating whether energy-saving features should be enabled or disabled for the network nodes. Specific nodes or regions within the network that require optimization may be included in the simulation scope. In an embodiment, the performance data may include at least one of throughput of the network nodes, latency of the network nodes, jitter of the network nodes, packet loss of the network nodes, data radio bearer (DRB) setup success rates of the network nodes, channel quality indicators (CQI) of the network nodes, signal-to-noise ratios (SNR) of the network nodes, and network energy consumption metrics of the network nodes, multi-domain context including external, alarm, and topology information which are aggregated and preprocessed by a unified Feature Fusion Layer before analysis, if energy-saving features are enabled. These metrics provide a comprehensive view of the network's performance and help identify areas for improvement.

[0123] Performance data is retrieved by the network configuration recommendation controller (124c) of the NDT server (124) from the management server (116) based on the plurality of network parameters. Advanced data processing capabilities are equipped in the NDT server (124) to handle large volumes of performance data efficiently. The management server (116) manages the network nodes by collecting and storing performance metrics from various network elements, ensuring data integrity and availability for analysis. Real-time performance of the network nodes is determined by the network configuration recommendation controller (124c) based on the retrieved performance data, fault and configuration data. This may include processing the data to generate real-time insights into the network's operational status and identifying any performance bottlenecks. If the real-time performance of the network node meets a performance threshold, the network configuration recommendation is received from an integrated CRE (110). Existing techniques are used by the CRE (110) to analyze the performance data and generate configuration recommendations that optimize network performance. The network configuration recommendation is transmitted to the MnS consumer (108), who may implement these recommendations to enhance network efficiency and ensure optimal service delivery.

[0124] In an embodiment, the MnS consumer (108) may include a memory (108a), a processor (108b), and a network configuration recommendation controller (108c) connected to the memory (108a) and the processor (108b). The network configuration recommendation controller (108c) is an innovative integrated circuit that includes a multi-core architecture, enabling dynamic adjustment of a location reporting trigger in a communication system. Each core is optimized for specific tasks related to NDT assisted network configuration generation based on performance monitoring recommendation in a communication network system.

[0125] The network configuration recommendation controller (108c) receives network configuration recommendations from the NDT server (124) and utilizes these recommendations for at least one of troubleshooting, performance analysis, congestion management, network failures, and suggesting corrective actions in the communication network system. Equipped with advanced algorithms, the network configuration recommendation controller (108c) analyzes real-time data from various network nodes to identify potential issues and optimize network performance. It may dynamically adjust network parameters such as bandwidth allocation, routing paths, and signal strength to mitigate congestion and enhance data throughput.

[0126] Further, the network configuration recommendation controller (108c) may predict network failures by monitoring key performance indicators and historical data, allowing for proactive maintenance and minimizing downtime.

[0127] The memory (108a or 124a) is configured to store instructions to be executed by the processor. The memory (108a or 124a) may include non-volatile storage elements. Examples of such non-volatile storage elements may include magnetic hard discs, optical discs, floppy discs, flash memories, or forms of electrically programmable memories (EPROM) or electrically erasable and programmable (EEPROM) memories. In addition, the memory may, in some examples, be considered a non-transitory storage medium. The term "non-transitory" may indicate that the storage medium is not embodied in a carrier wave or a propagated signal. However, the term "non-transitory" should not be interpreted that the memory (108a or 124a) is non-movable. In some examples, the memory (108a or 124a) may be configured to store larger amounts of information. In certain examples, a non-transitory storage medium may store data that may, over time, change (e.g., in random access memory (RAM) or cache).

[0128] The processor (108b or 124b) communicates with the memory, the communicator and the forbidden SNPN controller. The processor (108b or 124b) is configured to execute instructions stored in the memory and to perform various processes. The processor (108b or 124b) may include one or a plurality of processors, may be a general purpose processor, such as a central processing unit (CPU), an application processor (AP), or the like, a graphics-only processing unit such as a graphics processing unit (GPU), a visual processing unit (VPU), and / or an artificial intelligence (AI) dedicated processor such as a neural processing unit (NPU). Furthermore, the processor (108b or 124b) may include various processing circuitry and / or multiple processors. For example, as used herein, including the claims, the term "processor" may include various processing circuitry, including at least one processor, wherein one or more of at least one processor, individually and / or collectively in a distributed manner, may be configured to perform various functions described herein. As used herein, when "a processor", "at least one processor", and "one or more processors" are described as being configured to perform numerous functions, these terms cover situations, for example and without limitation, in which one processor performs some of recited functions and another processor(s) performs other of recited functions, and also situations in which a single processor may perform all recited functions. Additionally, the at least one processor may include a combination of processors performing various of the recited / disclosed functions, e.g., in a distributed manner. At least one processor may execute program instructions to achieve or perform various functions.

[0129] Fig. 4 illustrates a flow chart depicting a method for NDT assisted network configuration generation based on performance monitoring recommendation in a communication network system according to embodiments as disclosed herein. At operation S1, the NDT server (124) receives a request for simulation from the MnS consumer (108) associated with the live network (102). The request may comprise a plurality of network parameters associated with network nodes in the live network (102) to be simulated. In response to this request, the NDT server (124) acknowledges the MnS consumer (108). The NDT server (124) notifies the MnS consumer (108) that an NDT instance has been created and is ready to begin the simulation of the network nodes. The MnS consumer (108) may be a network operator or another service management function.

[0130] The NDT server (124) identifies the scope of the simulation. The scope of simulation specifies the target network nodes that need to be simulated. This may be provided in terms of target managed function (e.g., base stations, subnetworks) or in terms of a location. PM threshold is defined as the target thresholds for PMs, that is achieved by the resultant network configurations. Target QoS is defined as the target quality of service (e.g., LTE OCI or NR OFI tables) that is achieved by the resultant network configurations. Granularity is defined by the granularity to collect performance data, relating to the granularity period attribute as defined in TS 28622. Target Load is defined as the target load of the simulated network nodes, which may be defined in terms of virtual resource usage or in an implementation-specific way with a linear scale from 1 to 10. Target time is specified as the target time at which the performance of the nodes needs to be collected, defined in terms of time of day and relating to the scheduler IOC defined in TS 28622. Target energy saving is also specified, indicating whether the energy-saving features enabled or disabled for the network nodes being simulated. Additionally, the NDT server (124) may incorporate user-defined constraints such as maximum allowable latency or minimum throughput requirements to ensure the simulation aligns with specific operational goals.

[0131] At operation S2, the NDT server (124) retrieves performance data from the management server (116), fault and configuration data are also collected from a management server (116) based on the plurality of network parameters. The performance data for the NDT server (124) may comprise PM data (as defined in TS 28552 / 28554, e.g., handover parameters and TDD interference parameters), CM data (as defined in TS 28541 / 28622, e.g., handover parameters and TDD interference parameters), and alarm information and notifications (as per TS 28532). The management server (116) manages the network nodes. Furthermore, the NDT server (124) may retrieve weather data from the external weather system (120), which forecasts atmospheric temperature, atmospheric pressure, and vapor pressure for a given location. In an embodiment, the input data may include real-time and historical network performance metrics such as throughput, latency, packet loss, and quality of service (QoS) parameters as defined in 3GPP TS 28552. Configuration management information such as network slice resource usage, fault management data, and network topology configurations as defined in 3GPP TS 28554 may also be included. Additionally, the NDT server (124) may access user mobility patterns and device-specific performance metrics to enhance the accuracy of the simulation.

[0132] At operation S3, the NDT server (124) simulates the performance data and determines the real-time performance of the network nodes based on the retrieved performance data, fault and configuration data. The NDT server (124) is integrated with the CRE (110), which performs AI / ML-assisted simulations. The integrated CRE (110) uses unsupervised learning algorithms to detect outliers in performance data (e.g., sudden spikes in interference or latency) across the defined simulation nodes. Time series models within the CRE (110) are utilized to forecast the future performance of the network (102), helping predict issues such as throughput bottlenecks or high latency under varying load conditions. Deep learning techniques are used by the CRE (110) to analyze complex interactions between network parameters, enabling more accurate predictions and recommendations.

[0133] At operation S4, the NDT server (124) monitors the performance of the nodes in real-time by checking whether the real-time performance of the network node meets a performance threshold. Regular checks are conducted by the NDT server (124) to determine if any KPIs (e.g., latency, throughput, data radio bearer (DRB) setup success rates) exceed the given PM thresholds or QoS requirements. If any metric falls below the given threshold or QoS is compromised, the NDT server (124) sends a request to the CRE (110) to generate optimal network configuration recommendations. The machine learning models of CRE (110) include reinforcement learning to identify optimal configurations for base station parameters (antenna gain, transmission power, channel allocation). Heuristic-based tuning is applied to fine-tune specific configurations, such as reducing power to minimize interference or switching DRBs to optimize QoS flow retainability, resulting in restored degraded network performance. Genetic algorithms are leveraged by the CRE (110) to explore a wide range of configuration possibilities, ensuring that the effective solution is identified.

[0134] At operation S5, the method may include processing, by the NDT server (124), the collected multi domain telecom data using a spatio-temporal graph neural network (GNN) that incorporates a feature fusion attention mechanism, wherein the GNN is trained by a hybrid feedback loop operating on both real-world data and synthetic (simulation-generated) data states.

[0135] At operation S6, the method may include generating, by the NDT server (124), a network configuration recommendation for network automation, a recommendation may include an integrated explanation and confidence wherein value derived from the model's inference process.

[0136] At operation S7, the NDT server (124) receives optimal network configuration recommendations from the CRE (110). In an embodiment, the network configuration recommendation may comprise at least one of optimal network configuration values, predicted performance improvements of the network nodes, optimal PM or KPI values of the network nodes, and a configuration schedule specifying a timeframe for applying the network configuration recommendation. Specific actions such as adjusting the frequency bands used by the network nodes or implementing advanced scheduling algorithms to enhance overall network efficiency may also be included in the recommendations.

[0137] The NDT server (124) transmits the optimal network configuration recommendation to the MnS consumer (108) at operation S8. Additionally, the NDT server (124) sends a report including the final network configuration values recommendation (e.g., reconfiguring the gNB to improve DRB setup success rates or adjusting MIMO layers to enhance throughput), predicted performance improvements (e.g., lower latency, higher throughput, better energy efficiency), achieved PM / KPI values, and the timeframe in which specific configuration settings is applied. A detailed analysis of the expected impact on user experience and network stability may also be included in the report, providing the MnS consumer (108) with comprehensive insights for decision-making.

[0138] In an embodiment, the method may include, determining by the NDT server whether the real-time performance of the network node meets a performance threshold, receiving by the NDT server. processing, by the NDT server, the retrieved performance data, fault and configuration data using a spatio-temporal graph neural network (STGNN) that incorporates a feature fusion attention mechanism, further the STGNN is trained by a hybrid feedback loop operating on both real-world data and synthetic (simulation-generated) data states; generating, by the NDT server, a network configuration recommendation for network automation, wherein said recommendation may include an integrated explanation and confidence value derived from the model's inference process.

[0139] In an embodiment, the method may include the plurality of network parameters associated with network nodes may comprise at least one of: a simulation scope defining target network nodes to be simulated; a performance management threshold defining target thresholds for performance metrics (PMs) to be achieved by the network configuration recommendation; a target quality of service (QoS) defining quality of service parameters for the network configuration recommendation; a granularity defining a granularity period for collecting the performance data; a target load defining a target load of the network nodes to be simulated; a target time defining a time at which the performance of the network nodes needs to be collected; and a target energy-saving parameter indicating whether energy saving features will be enabled or disabled for the network nodes.

[0140] In an embodiment, the method may include the performance data may comprise at least one of throughput of the network nodes, latency of the network nodes, jitter of the network nodes, packet loss of the network nodes, DRB setup success rates of the network nodes, channel quality indicators (CQI) of the network nodes, signal-to noise ratios (SNR) of the network nodes, and network energy consumption metrics of the network nodes, multi-domain context including external, alarm, and topology information which are aggregated and preprocessed by a unified Feature Fusion Layer before analysis, if energy-saving features are enabled.

[0141] In an embodiment, the method may include the network configuration recommendation may comprise at least one of optimal network configuration values, predicted performance improvements of the network nodes, optimal PM or KPI values of the network nodes, and a configuration schedule specifying a timeframe for applying the network configuration recommendation.

[0142] In an embodiment, the method may include generating, by the NDT server (124), the network configuration recommendation may comprise: detecting, by the NDT server (124), anomalies in the performance data across the network nodes using an unsupervised learning model, wherein the anomalies include sudden spikes in interference or latency and further predicting, by the NDT server (124), a future network performance of the network nodes using a time-series model; adjusting, by the NDT sever (124), at least one of a transmission powers of the network nodes, a bandwidth allocation of the network nodes, and an antenna configuration of the network nodes based on the detected anomalies and future network performance of the network nodes and furthermore, identifying, by the NDT server (124), mitigate interference from neighboring cells based on historical data and specified interference thresholds; and generating, by the NDT server (124), the network configuration recommendation based on the mitigate interference and at least one of the adjusted transmission powers of the network nodes, the adjusted bandwidth allocation of the network nodes, and the adjusted antenna configuration of the network nodes.

[0143] In an embodiment, the method may include determining, by the NDT server (124), the real-time performance of the network nodes based on the retrieved performance data, fault and configuration data may include simulating, by the NDT server (124), the network nodes by using machine learning and heuristic models by applying an artificial intelligence model on the performance data retrieved from the Management server (116), further the machine learning models are dynamically updated using a hybrid training feedback loop integrating both simulation and real-world performance feedback and determining, by the NDT server (124), the real-time performance of the network node the real-time performance monitoring of the network nodes based on the performance parameters.

[0144] In an embodiment, the method may include receiving, by the MnS consumer (108), the network configuration recommendations from the NDT server (124) and utilizing, by the MnS consumer (108), the network configuration recommendations for at least one of troubleshooting, performance analysis, wherein simulations and predictive modeling are performed by the spatio-temporal GNN architecture, wherein all recommendations are accompanied by machine-extracted explanations and confidence scores generated within the network digital twin system, congestion management, network failures and suggesting corrective actions in the communication network system.

[0145] In an embodiment, the method may include transmitting, by the NDT server (124), an NDT creation or activation notification message to the MnS Consumer (108) indicating that the NDT server (124) is ready to begin simulation of the network nodes.

[0146] Fig. 5 illustrates a flow chart indicating operations followed by the MnS consumer in a communication network system according to embodiments as disclosed herein. At operation (S9), the MnS consumer (108) receives the generated optimal network configuration recommendation from the NDT server (124). The recommendation is based on real-time data analysis and predictive algorithms that consider various network parameters such as bandwidth usage, latency, and error rates. At operation (S10), the MnS consumer (108) utilizes this optimal network configuration recommendation received at operation S9 for at least one of troubleshooting, performance analysis, congestion management, network failures, and suggesting corrective actions in the communication network system. The MnS consumer (108) may employ machine learning models to monitor network performance and dynamically adjust configurations to maintain optimal operation.

[0147] At operation S10, the MnS consumer (108) utilizes the network configuration recommendations in the communication network system. The MnS consumer (108) implements the network configuration recommendation on the live network (102) either manually or automatically, without the NDT server (124) exerting direct control over the infrastructure of the live network (102). Automated scripts or orchestration tools may be used by the MnS consumer (108) to apply the recommended configurations, ensuring a seamless transition and minimal disruption to ongoing network operations.

[0148] Fig. 6 illustrates about sequence diagram, a flow of sequences performed to obtain network configuration recommendations according to embodiments as disclosed herein. In an embodiment, the network configuration recommendation by the NDT server (124) may comprise detecting anomalies in the performance data across the network nodes using an unsupervised learning model. The unsupervised learning model may include clustering algorithms such as K-means or DBSCAN to identify patterns and deviations in the network performance metrics. Predicting a future network performance of the network nodes using a time-series model may include utilizing models like ARIMA or LSTM to forecast traffic loads and latency trends. Adjusting at least one of a transmission power of the network nodes, a bandwidth allocation of the network nodes, and an antenna configuration of the network nodes based on the detected anomalies and future network performance of the network nodes ensures optimal resource utilization and minimizes congestion. Identifying and mitigating interference from neighboring cells based on historical data and specified interference thresholds may include analyzing past interference events and setting thresholds using statistical methods. Generating the network configuration recommendation based on the mitigated interference and at least one of the adjusted transmission powers of the network nodes, the adjusted bandwidth allocation of the network nodes, and the adjusted antenna configuration of the network nodes ensures a balanced and efficient network operation.

[0149] In an embodiment, the NDT server (124) generates the network configuration recommendation to mitigate interference and at least one of the adjusted transmission powers of the network nodes, the adjusted bandwidth allocation of the network nodes, and the adjusted antenna configuration of the network nodes. The NDT server (124) utilizes advanced algorithms to dynamically adjust these parameters in real-time, ensuring that the network adapts to changing conditions and maintains optimal performance. The NDT server (124) generates configuration outputs for network optimization including but not limited to bandwidth allocation, load balancing, energy efficiency configurations, and traffic management policies. These outputs are derived from comprehensive data analysis and predictive modeling, allowing for proactive adjustments that enhance network reliability and user experience. Additionally, the server may incorporate machine learning techniques to improve its recommendations based on feedback and evolving network conditions.

[0150] Fig. 6 illustrates about sequence diagram, flow of sequences performed by "MnS Producer (NDT)" block to obtain network configuration recommendations.

[0151] As illustrated in Fig. 6, it is about a communication sequence between different components involved in the ML-based NDT simulation and configuration recommendation process, includes MTS Consumer (108), MTS producer (125), a system incorporating a novel ML-based ST-GRM with feature fusion and hybrid simulation trainer, management system (116) including IPA MTS producer, ICM MTS producer, and EFM MTS producer, external weather system (120), configuration recommendation engine (110) such as 3GPP MDAS.

[0152] At operation 601, the MTS producer (125), upon receiving the request, sends an NDT creation / activation notification (operation 603) and initiates data collection by fetching performance management (PM), configuration management (CM), and alarm data (operation 602a) from the management system (116), and weather data (operation 602b) from the external weather system (120).

[0153] At operation 604, the MTS producer (125), which includes a machine learning model (ML ST-GRM) integrated with a feature fusion layer and a hybrid simulation trainer, performs simulation processing. This may include monitoring of network nodes and executing hybrid simulation-learning loops to replicate and analyze the network's operational behavior. At operation 605, the system performs performance degradation detection based on QoS KPI predictions using trained ML models. At operation 606, a configuration recommendation is generated using explainable ML techniques with associated confidence scores. The generated configuration data is forwarded to the configuration recommendation engine (110), which includes a system such as a 3GPP MDAS, for further processing or implementation.

[0154] At operation 607, the MTS producer (125) reports the generated configuration recommendations back to the MTS consumer (108). These recommendations include explainable machine learning outputs along with confidence levels, enabling informed decision-making. The input data used by the NDT simulation may include PM data (e.g., TS 28.552 / 28.554), CM data (TS 28.552 / 28.554), alarm data, topology reference parameters, and derived features (e.g., delta CM and delta KPIs), as well as weather data. All these inputs are passed through a Feature Fusion Layer prior to ML processing to enhance the accuracy and contextual relevance of the simulation outcomes. This integrated approach ensures data-driven, reliable, and scalable network configuration recommendations.

[0155] Fig. 7 illustrates prediction of TDD interference in a communication network according to embodiments as disclosed herein. Digital twins are virtual representations of physical systems, including base stations and the communication channels, that are used to simulate, predict, and optimize real-world processes. These digital twins may model complex interactions between various network elements, taking into account factors such as signal propagation, interference patterns, and network load. Input parameters are received from a real network (102). The input parameters include parameters related to the base station like unique aggressor ID, unique victim ID, geographical and symbol distance between aggressor and victim cell, interference power, SINR, mitigation action taken, average interference power, duration of interference, total interference power, antenna type, antenna height, antenna azimuth, and transmit power, etc. These parameters are used for accurately modeling the interference scenarios and predicting the impact on network performance. Parameters related to weather, like atmospheric temperature, atmospheric pressure, transmit power, atmospheric temperature, atmospheric pressure, vapor pressure, etc., are obtained from the weather system (120). Weather conditions may significantly affect signal propagation and interference, making it used to incorporate these parameters into the simulation. These parameters are provided to the NDT server (124) based digital twin solution. The NDT server (124) generates output data including aggressor-victim pair and their IDs, timestamp, and mitigation solution. The server uses advanced algorithms to analyze the input data and generate predictions about potential interference issues and recommended mitigation actions. This output data is applied to simulated network scenarios (128). Further, this simulated work scenario may help in reducing TDD interference. By testing various configurations and mitigation strategies in the simulated environment, network operators may identify the effective solutions to minimize interference and optimize network performance.

[0156] Fig. 8 illustrates the NDT based predictive maintenance of hardware in accordance with an embodiment of the disclosure. The NDT server (124) may comprise of a unified data repository (130) and a model (132). The unified data repository (130) is designed to store large volumes of real-time data, including performance metrics, error logs, and environmental conditions, ensuring comprehensive data collection for accurate analysis. A real-time data is collected from the live network (102) and stored in the unified data repository (130). The data collection process may include various sensors and monitoring tools strategically placed within the network infrastructure to capture relevant data points. The AI model (132) receives this real-time data from the unified data repository (130). The AI model (132) is equipped with advanced machine learning algorithms capable of processing and analyzing the data to identify patterns and anomalies indicative of potential hardware failures.

[0157] The data is applied on a mirror model (132). The mirror model (132) is a virtual representation of the live network (102), replicating its configuration and operational parameters to simulate real-world conditions accurately. A set of hardware attributes / features related to the live network (102) are applied on an AI / ML model to predict any chances of service failure. These attributes / features include temperature variations, power consumption levels, signal strength, and historical failure rates, which are used for predicting hardware reliability. A set of corrective measures are implemented on a simulation model. The simulation model tests various corrective actions, such as adjusting network configurations, rerouting traffic, or replacing faulty components, to determine their effectiveness in preventing service failures. After this, the simulated output is checked; if this output is improved, the corrective measures are implemented on the live network (102). The improvement is measured based on specified performance metrics, such as reduced error rates, enhanced signal quality, and increased uptime.

[0158] If the output of the simulated model does not show any improvement, another set of hardware attributes / features are applied on the AI / ML model to predict any chances of service failure. This iterative process ensures that all possible scenarios and corrective actions are thoroughly evaluated to find the effective solution. This operation is repeated till the output of the simulated model does not show any improvement. The continuous loop of testing and evaluation helps in refining the AI / ML model's accuracy and reliability in predicting hardware failures. At this stage, the corrective measures are sent to the live network (102) through a control signal. The control signal is transmitted via secure communication channels to ensure the integrity and authenticity of the corrective instructions, minimizing the risk of unauthorized access or tampering.

[0159] Input parameters:

[0160]

[0161]

[0162] Fig. 9 illustrates the digital twin-based predictive maintenance of hardware according to embodiments as disclosed herein. A production environment analyzer framework (PEaF) (134) may be used to address any limitations in a continuous deployment (CD) process for workflow lifecycle management of the live network (102). The PEaF (134) is designed to operate in real-time, monitoring the production environment to ensure optimal performance. It collects raw data using a data collector (136), which interfaces with various sensors and monitoring tools deployed across the network infrastructure. The collected data is pre-processed to remove noise and irrelevant information before being fed into the clustering algorithm. The framework applies K-Means clustering to group similar data points and assigns scores to each cluster based on specified criteria such as performance metrics, error rates, and usage patterns, and stores it in a cluster database (138).

[0163] The PEaF (134) also may comprise a process analyzer (140) which uses AI / ML techniques to analyze the production environment and predict the health status of hardware components. The process analyzer (140) employs advanced machine learning models, such as neural networks and decision trees, to identify patterns and anomalies in the data that may indicate potential hardware failures. It learns from historical data and adapts its predictions to improve accuracy over time. A process analyzer (140) analyzes the production environment and excludes deteriorating hardware from the CD. This exclusion process may include generating alerts and recommendations for maintenance or replacement of the identified hardware components, thereby preventing them from being included in future deployments. The analyzed data and output from a decision-making scheduler (142) are stored in a classified database (144). The decision-making scheduler (142) uses complex algorithms to prioritize maintenance tasks and schedule them in a way that minimizes disruption to the production environment. The data from this classified database (144) may be used for reducing service failures considerably. This ensures reliability and stability of the production environment before deploying / upgrading services, as it allows for proactive maintenance and timely interventions based on predictive insights.

[0164] The input data used is as illustrated in the below table:

[0165]

[0166] The ideal configuration for maximizing 5G performance may include several cases:

[0167] Case 1 may include analyzing network topology. Understanding the physical layout of the network, including the locations of base stations, antennas, and backhaul links, isessentialfor optimizing performance.

[0168] Case 2 requires assessing the channel quality index (CQI). The choice of modulation and coding scheme depends on various factors such as the characteristics of the channel, therequireddata rate, and the desired level of error correction. Based on the CQI, which represents the signal strength and noise level of the channel, different modulation and coding schemes (MCS) may be selected.

[0169] Case 3 evaluates network energy consumption. Energy consumption in networks is amajorexpense for operators, particularly in developed markets where it accounts for 10-15% of total network operating expenses. In developing markets with a high number of remote sites, energy costs may even reach up to 50%.

[0170] Case 4 focuses on NDT-based call drop optimization using handover (HO) statistics and geo-coordinates. This case may include generating an optimum configuration for a smooth handover process.

[0171] Case 5 addresses NDT-based time division duplex (TDD) interference management. Prediction of potential TDD interference may include identifying aggressor base stations along with their victim base stations using NDT.

[0172] Case 6 pertains to NDT-based hardware predictive maintenance. Potential hardware issues may be predicted using NDT-supported simulation.

[0173] The new entity location management function (LMF) is central in the 5G positioning architecture. Measurements and assistance information are received by the LMF from the next generation radio access network (NG-RAN) and the mobile device, otherwise known as the user equipment (UE), via the access and mobility management function (AMF) over the NLs interface to compute the position of the UE. To derive the best configuration management (CM) attributes in the LMF, the following performance management (PM) attributes threshold shall be set.

[0174] Antennas at the receiver enable precise angle of arrival (AOA) measurements in uplink, utilizing a range of positioning methods including observed time difference of arrival (OTDOA), uplink time difference of arrival (UL-TDOA), and power-based positioning methods (TR 28552 Section 511332), round trip time (RTT), and angle-based positioning.

[0175] The SRS parameters define the characteristics of the reference signal, such as periodicity, duration, and bandwidth. Optimizing these parameters may enhance the accuracy of channel measurements performed by the LMF. Allocating resources for the LMF, such as processing power and memory, is used for ensuring timely and accurate channel measurements.

[0176] Assessing the CQI is used to ensure data reliability in wireless communication systems. Modulation and coding scheme selection plays an important role in achieving reliable data transmission over a given channel. Various factors influence the choice of modulation and coding scheme, including the characteristics of the channel, the required data rate, and the level of error correction. The CQI, which represents the signal strength and noise level of the channel, serves as a basis for selecting different modulation and coding schemes (MCS).

[0177] A CQI PM threshold is established to derive the best MCS CM value. This measurement provides the distribution of wideband channel quality indicator (CQI) reported by UEs in the cell, as detailed in TR 28552 Section 51111. When the CQI value is high, higher order MCS like 64QAM or 256QAM may be utilized along with lower code rates to increase data throughput. Conversely, if the CQI value is low, indicating poor channel conditions with weak signals and high interference, lower order modulations like QPSK or BPSK should be used with higher code rates for better error correction.

[0178] NDT-assisted PM helps in choosing the appropriate combination of modulation and coding scheme based on the current channel quality index. This approach maximizes data throughput while maintaining acceptable error rates.

[0179] In another case evaluating network energy consumption, 5G will introduce new network features that significantly impact energy consumption and efficiency. Massive MIMO and antenna beam steering will enhance link budgets and compensate for fading, especially with millimeter-wave carrier frequencies. While these improvements boost spectral efficiency per area, they may also increase power consumption. The following PM threshold is set to derive the CM value: PM attributes threshold. To establish metrics and KPIs for mobile network efficiency:

[0180] 1. Energy per bit: In urban environments where network planning is typically capacity-constrained, the concept of energy per bit is frequently used. Energy per bit is represented by the symbol E and signifies the amount of energy consumed during a specified observation period measured at the medium access control (MAC) layer.

[0181] 2. Power per area unit: In suburban or rural environments where network planning is predominantly determined by coverage requirements, alternative metrics are often used. These metrics focus on the achieved coverage area rather than energy consumption per bit. Despite their popularity among academic institutions, the inverse measures of these metrics are generally preferred for reporting and evaluating products.

[0182] 3. Energy per bit: The number of delivered bits per unit of energy expressed as [bit / J] is a common metric used to assess equipment and operational network energy efficiency in various environmental standards established by organizations such as ITU-T and ETSI.

[0183] 4. Coverage area per daily energy consumption: The energy efficiency parameter for operational mobile networks is commonly expressed as [m2 / J] or [m2 / Wh].

[0184] Meanwhile, the previously mentioned metrics are also frequently utilized in simulated scenarios to estimate energy efficiency without necessitating real measurements of the included parameters. Best CM attributes: NDT will develop a BS model, describing the power consumption of varying configurations: number of sectors, transmitters per sector, maximum installed RF power, actual load.

[0185] In another case, the NDT based call drop optimization using HO statistics and geo-coordinates / optimization of cell selection and reselection parameters in mobile networks. By monitoring the handover parameters, issues related to data loss over the radio link between the base station and the user equipment may be analysed. Optimizing the parameters like individual cell offset, hysteresis, time-to-trigger, threshold, and time-to-trigger may help improve the reliability of the radio link and enhance the quality of service experienced by users. This optimization may lead to reduced dropped calls, improved connection stability, and better overall network performance.

[0186] The table given below indicates the input PM parameters for monitoring (Ref: 3GPP TS 28.552).

[0187]

[0188] The table given below indicates the output configurable parameters for optimization (Ref: 3GPP TS 38.331).

[0189]

[0190] It will be appreciated that various embodiments of the disclosure according to the claims and description in the specification can be realized in the form of hardware, software or a combination of hardware and software.

[0191] Any such software may be stored in non-transitory computer readable storage media. The non-transitory computer readable storage media store one or more computer programs (software modules), the one or more computer programs include computer-executable instructions that, when executed by one or more processors of an electronic device, cause the electronic device to perform a method of the disclosure.

[0192] Any such software may be stored in the form of volatile or non-volatile storage, such as, for example, a storage device like read only memory (ROM), whether erasable or rewritable or not, or in the form of memory, such as, for example, random access memory (RAM), memory chips, device or integrated circuits or on an optically or magnetically readable medium, such as, for example, a compact disk (CD), digital versatile disc (DVD), magnetic disk or magnetic tape or the like. It will be appreciated that the storage devices and storage media are various embodiments of non-transitory machine-readable storage that are suitable for storing a computer program or computer programs comprising instructions that, when executed, implement various embodiments of the disclosure. Accordingly, various embodiments provide a program comprising code for implementing apparatus or a method as claimed in any one of the claims of this specification and a non-transitory machine-readable storage storing such a program.

[0193] While the disclosure has been shown and described with reference to various embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the disclosure as defined by the appended claims and their equivalents.

Claims

1.A method for network digital twin (NDT) assisted network configuration generation based on performance monitoring in a communication network system, the method comprising:receiving, by a NDT server (124), a request for simulation from a management service (MnS) consumer (108) associated with a live network (102), wherein the request comprises a plurality of network parameters associated with network nodes in the live network (102) to be simulated;retrieving, by the NDT server (124), performance data from a management server (116) based on the plurality of network parameters, wherein the management server (116) manages the network nodes;determining, by the NDT server (124), a performance of the network nodes based on the retrieved performance data, fault and configuration data;receiving, by the NDT server (124), a network configuration recommendation from a configuration recommendation engine (CRE) (110), wherein the performance of the network node meets a performance threshold; andtransmitting, by the NDT server (124), the network configuration recommendation to the MnS consumer (108).2.The method of claim 1, wherein the plurality of network parameters associated with network nodes comprises at least one of:a simulation scope defining target network nodes to be simulated;a performance management threshold defining target thresholds for performance metrics (PMs) to be achieved by the network configuration recommendation;a target quality of service (QoS) defining quality of service parameters for the network configuration recommendation;a granularity defining a granularity period for collecting the performance data;a target load defining a target load of the network nodes to be simulated;a target time defining a time at which the performance of the network nodes needs to be collected; anda target energy-saving parameter indicating whether energy-saving features are enabled or disabled for the network nodes.3.The method of claim 1, wherein the determining, by the NDT server (124), the performance of the network nodes comprising determining whether the performance of the network node meets the performance threshold, wherein the performance threshold includes target thresholds for performance metrics (PMs) achieved by the network configuration recommendation.4.The method of claim 1, wherein the retrieved performance data, fault and configuration data are processed using a spatio-temporal graph neural network (STGNN) by a feature fusion attention mechanism, wherein the STGNN is trained by a hybrid feedback loop operating on both real-world data and simulation-generated data states.5.The method of claim 1, wherein the network configuration recommendation includes an integrated explanation and confidence value derived from the model's inference process, andwherein the network configuration recommendation comprises at least one of network configuration values, predicted performance improvements of the network nodes, performance management (PM) or key performance indicator (KPI) values of the network nodes, and a configuration schedule specifying a timeframe for applying the network configuration recommendation.6.The method of claim 1, wherein the performance data comprises at least one of throughput of the network nodes, latency of the network nodes, jitter of the network nodes, packet loss of the network nodes, data radio bearer (DRB) setup success rates of the network nodes, channel quality indicators (CQI) of the network nodes, signal-to-noise ratios (SNR) of the network nodes, and network energy consumption metrics of the network nodes, multi-domain context including external, alarm, and topology information which are aggregated and preprocessed by a unified feature fusion layer before analysis, wherein energy-saving features are enabled.7.The method of claim 1,further comprising:detecting, by the NDT server (124), anomalies in the performance data across the network nodes using an unsupervised learning model, wherein the anomalies include sudden spikes in interference or latency;predicting, by the NDT server (124), a future network performance of the network nodes using a time-series model;adjusting, by the NDT sever (124), at least one of a transmission power of the network nodes, a bandwidth allocation of the network nodes, and an antenna configuration of the network nodes based on the detected anomalies and future network performance of the network nodes;identifying, by the NDT server (124), mitigate interference from neighboring cells based on historical data and interference thresholds; andgenerating, by the NDT server (124), the network configuration recommendation based on the mitigate interference and at least one of the adjusted transmission powers of the network nodes, the adjusted bandwidth allocation of the network nodes, and the adjusted antenna configuration of the network nodes.8.The method of claim 1, determining, by the NDT server (124), the performance of the network nodes based on the retrieved performance data, fault and configuration data comprises:simulating, by the NDT server (124), the network nodes by using machine learning and heuristic models by applying an artificial intelligence model on the performance data retrieved from the management server (116), wherein the machine learning models are dynamically updated using a hybrid training feedback loop integrating both simulation and real-world performance feedback; anddetermining, by the NDT server (124), the performance of the network node the performance monitoring of the network nodes based on the performance parameters.9.The method of claim 1, wherein the network configuration recommendation is utilized for at least one of troubleshooting, performance analysis, simulation and predictive modeling,wherein the simulations and predictive modeling are performed by a spatio-temporal graph neural network (STGNN) architecture, andwherein the network configuration recommendation is accompanied by machine-extracted explanations and confidence scores generated within the network digital twin system, congestion management, network failures and suggesting corrective actions in the communication network system.10.The method of claim 1, further comprising transmitting, by the NDT server (124), an NDT creation or activation notification message to the MnS consumer (108) indicating that the NDT server (124) is ready to begin simulation of the network nodes.11.A method for generating dynamic network configuration recommendations using a network digital twin (NDT) server (125) in a communication network system, the method comprising:receiving, by the NDT server (125), a simulation request from a management service (MnS) consumer (108) associated with a live communication network (102), wherein the request comprises a plurality of network parameters associated with network nodes;retrieving, by the NDT server (125), network-related data including performance data from a performance assurance system (116a), fault data from a fault management system (117), and configuration data from a configuration management system (118), each of which is part of a management server (116);obtaining, by the NDT server (125), external environmental data from an external system (120);generating, by the NDT server (125), fused feature embeddings by processing the retrieved data using domain-specific encoders and a scenario-aware gating mechanism;constructing, by the NDT server (125), a spatio-temporal graph of network elements using a spatio-temporal graph neural network (STGNN) inference engine (125c), and applying message-passing operations to predict network states and generate configuration recommendations;training, by the NDT server (125), the STGNN inference engine (125c) using a hybrid learning loop incorporating real network data and validated synthetic data;evaluating, by the NDT server (125), the generated recommendations to determine feasibility, compliance, and confidence score; andtransmitting, by the NDT server (125), configuration recommendations to the MnS consumer (108) for deployment in the live communication network (102).12.The method of claim 11, wherein the retrieved data is processed by a feature fusion layer (125b), wherein the feature fusion layer (125b) applies temporal alignment using one or more of temporal convolutions or recurrent neural units to synchronize heterogeneous domain data arriving at different timescales, andwherein the STGNN inference engine (125c) dynamically constructs or deletes edges in the spatio-temporal graph based on real-time triggers such as key performance indicators (KPIs), alarm signals, interference, or energy metrics.13.The method of claim 12, wherein a scenario policy controller (125a) supervises the feature fusion layer (125b) and the STGNN inference engine (125c) by dynamically adjusting fusion weights and selecting message-passing policies based on detected network scenarios,wherein the STGNN inference engine (236c) is trained using a simulation feedback trainer (125e), wherein the simulation feedback trainer (125e) filters low-quality synthetic events and prioritizes rare but critical network events to fine-tune the STGNN inference engine (125c) in near real-time, andwherein the configuration recommendations is evaluated by a constraints and confidence gate (125f), wherein the constraints and confidence gate (125f) detects a constraint projection confidence score for each recommendation, and selects only those that meet predefined performance, risk, and policy criteria for final deployment.14.A network digital twin (NDT) server for NDT assisted network configuration generation based on performance monitoring, the NDT server comprises:memory (124a) storing instructions; andat least one processor (124b) communicatively coupled to the memory (124a),wherein the instructions, when executed by the at least one processor individually or collectively, cause the NDT server to:receive a request for simulation from a management service (MnS) consumer (108) operably connected to a live network (102), wherein the request comprises a plurality of network parameters associated with network nodes in the live network (102) to be simulated;retrieve performance data from a management server (116) based on the plurality of network parameters, wherein the management server (116) manages the network nodes;determine a performance of the network nodes based on the retrieved performance data, fault and configuration data;receives a network configuration recommendation from a configuration recommendation engine (CRE) (110), wherein the performance of the network node meets a performance threshold; andtransmit the network configuration recommendation to the MnS consumer (108).15.A network digital twin (NDT) server (125) for generating dynamic network configuration recommendations in a communication network system, the NDT server comprising:memory (124a) storing instructions; andat least one processor (124b) communicatively coupled to the memory (124a),wherein the instructions, when executed by the at least one processor individually or collectively, cause the NDT server to:receive a simulation request from a management service (MnS) consumer (108) associated with a live communication network (102), wherein the request comprises a plurality of network parameters associated with network nodes;retrieve, network-related data including performance data from a performance assurance system (116a), fault data from a fault management system (117), and configuration data from a configuration management system (118), each of which is part of a management server (116);obtain, external environmental data from an external system (120);generate fused feature embeddings by processing the retrieved data using domain-specific encoders and a scenario-aware gating mechanism;constructs, spatio-temporal graph of network elements using a spatio-temporal graph neural network (STGNN) inference engine (125c), and apply message-passing operations to predict network states and generate configuration recommendations;train the STGNN inference engine (125c) using a hybrid learning loop incorporating real network data and validated synthetic data;evaluates the generated recommendations to determine feasibility, compliance, and confidence score; andtransmit configuration recommendations to the MnS consumer (108) for deployment in the live communication network (102).

Images