AI-Powered Network Resource Management System Based on Dynamic Carbon Awareness

Abstract

The invention relates to a Multi-Agent Artificial Intelligence-Based Autonomous Network Management System and Method that enables autonomous network management through the coordinated operation of artificial intelligence agents with different areas of expertise in communication networks.The system includes the Network Data Collection Module (1), which collects operational data of the communication network; the Data Preprocessing and Distribution Module (2), which prepares the data for analysis; Artificial Intelligence Agents (3), which perform independent evaluations for different optimization goals; the Agent Coordination Module (4), which ensures coordination between agents; the Dynamic Decision Consensus Module (5), which evaluates decision proposals together and forms a joint decision; the Conflict Resolution Module (6), which manages decision differences; the Autonomous Network Decision Module (7), which converts the formed joint decision into an implementable network management decision; the Network Management and Control Module (8), which applies the decisions to the communication network; the Performance Monitoring and Feedback Module (9), which continuously monitors system performance; and the Learning and Model Update Module (10), which improves the decision-making processes of artificial intelligence agents using feedback information.Thanks to this invention, decision proposals generated by different artificial intelligence agents are dynamically reconciled, decision conflicts are effectively managed, autonomous decision-making mechanisms that adapt to changing network conditions are created, and communication networks are managed in a more reliable, flexible, scalable, and efficient manner.

Description

1 TARIFF AI-Powered Network Resource Management Based on Dynamic Carbon Awareness The system Technical Area This invention enables autonomous management of communication networks, artificial intelligence-based decision support systems, Multi-agent artificial intelligence, distributed artificial intelligence architectures, network It relates to the technical fields of optimization and network automation. More specifically, the invention; fifth 10 5th generation (5G), fifth generation beyond (5G-Advanced), sixth generation (6G) and next generation communications radio access network (RAN), core network, transport network used in networks (Transport Network), Software Defined Networking (SDN), network slicing Network slicing, open radio access network (Open RAN) and similar communication infrastructures Multi-agent artificial intelligence for network management and resource optimization processes It relates to an intelligence-based system and method. Within the scope of the invention; traffic density, network load, and service obtained from communication networks. Quality of life (QoS) indicators include energy consumption, safety events, user activity, and network performance. Performance metrics and similar operational data are generated by artificial intelligence (AI) with different areas of expertise. The data is being analyzed independently by the intelligence agents, with each AI agent analyzing its own data. It generates decision recommendations related to its area of ​​expertise. The generated decision recommendations are then... The existing communication network is evaluated within a common decision-making mechanism. This ensures that the most appropriate management decision is determined for the situation. The invention is particularly relevant for self-managing networks, which are self-optimizing. with Self-Optimizing Networks (SON) and AI-Native network architectures Multi-agent decision-making that can be used in autonomous network structures. It is aimed at mechanisms that reduce human intervention in network management processes, different being able to evaluate optimization goals together and adapt them to changing network conditions in real time. aiming to implement autonomous decision-making processes that can adapt over time. It includes systems and methods. 35 2 State of the Art Today, we have fifth generation (5G), fifth generation beyond (5G-Advanced) and sixth generation (6G) With the proliferation of communication networks, the complexity of communication infrastructures has become significant. It has increased considerably. Radio access network (RAN), core network, transport network (Transport 5 Network slicing, software-defined networking - Simultaneously managing multi-layered structures such as SDN and cloud-based network architectures, This leads to the inadequacy of traditional network management methods. Therefore, the latest In recent years, artificial intelligence-based decision support systems and autonomous networks have been used in network management processes. The use of management approaches is becoming increasingly common. 10 Current AI-based network management systems mostly use a single AI model. Network traffic forecasting, resource allocation, energy optimization, and quality of service are all possible using this method. A specific optimization problem, such as improvement or security analysis, is being solved. This The artificial intelligence model used in the systems evaluates different network indicators together to create a single 15 making a decision and performing network management operations in accordance with that decision. is being carried out. However, this needs to be considered simultaneously in modern communication networks. The number of optimization goals is constantly increasing. For example, reducing energy consumption is 20 The goal of improving service quality and security can conflict with each other. These requirements can increase network latency, and balancing traffic density requires energy. This can negatively affect productivity. Similarly, operator policies, user Different criteria such as mobility, network load, resource utilization rates, and security risks are also the same. These are among the parameters that need to be considered at the moment. 25 A single AI model achieving so many different optimization goals simultaneously. The assessment suggests that the increasing complexity of the model reduces the transparency of decision-making processes. to reduce, effectively prioritize between different optimization goals 30 This can be the reason. Furthermore, existing solutions involve independent professionals with different areas of expertise. for decision-making mechanisms to work in a coordinated manner in order to reach a common decision There is no integrated structure. In recent years, multi-agent artificial intelligence (AI) architectures have evolved into 35 different types. Although they have begun to be used in various application areas, these approaches are mostly independent. 3 They are designed based on the principle of task sharing or cooperation; in communication networks. Dynamic evaluation of numerous optimization objectives that may conflict with each other, eliminating potential differences in decision-making among agents and establishing a single system for the entire entity. Comprehensive decision coordination for the creation of autonomous network management decisions. It does not offer a mechanism. 5 For these reasons, the communication network of artificial intelligence agents with different areas of expertise They can independently evaluate the data obtained within their own areas of responsibility, The decision proposals generated by the agents are dynamic within a common decision-making process. This analysis shows that the priorities between different optimization goals are determined by the network's instantaneous 10 It can be managed according to the situation, and as a result, the communication network becomes more reliable and flexible. A new system and method is needed that allows it to be managed in an explainable and autonomous way. It is heard. Purpose of the Invention 15 Today, we have fifth generation (5G), fifth generation beyond (5G-Advanced) and sixth generation (6G) With the proliferation of communication networks, the complexity of communication infrastructures has become significant. It has increased considerably. Radio access network (RAN), core network, transport network Networking, network slicing, software-defined networking (Software Defined Networking 20 - Simultaneously managing multi-layered structures such as SDN and cloud-based network architectures, This leads to the inadequacy of traditional network management methods. Therefore, the latest In recent years, artificial intelligence-based decision support systems and autonomous networks have been used in network management processes. The use of management approaches is becoming increasingly common. Current AI-based network management systems mostly use a single AI model. Network traffic forecasting, resource allocation, energy optimization, and quality of service are all possible using this method. A specific optimization problem, such as improvement or security analysis, is being solved. This The artificial intelligence model used in the systems evaluates different network indicators together to create a single system. making a decision and performing network management operations in accordance with the decision obtained 30 is being carried out. However, this needs to be considered simultaneously in modern communication networks. The number of optimization goals is constantly increasing. For example, reducing energy consumption. The goal of improving service quality and the goal of security can conflict with each other. 35 These requirements can increase network latency, and balancing traffic density requires energy. 4 This can negatively affect productivity. Similarly, operator policies, user Different criteria such as mobility, network load, resource utilization rates, and security risks are also the same. These are among the parameters that need to be considered at this time. A single AI model can achieve so many different optimization goals simultaneously. The assessment suggests that the increasing complexity of the model reduces the transparency of decision-making processes. to reduce, effectively prioritize between different optimization goals unmanageability and limited ability to adapt to changing network conditions This can be the reason. Furthermore, existing solutions involve independent professionals with different areas of expertise. 10. This refers to the coordinated work of decision-making mechanisms to reach a common decision. There is no integrated structure. In recent years, multi-agent artificial intelligence (AI) architectures have evolved in different ways. Although they have begun to be used in various application areas, these approaches are mostly independent. They are designed based on task sharing or cooperation; in communication networks 15 Dynamic evaluation of numerous optimization objectives that may conflict with each other, eliminating potential differences in decision-making among agents and establishing a single system for the entire entity. Comprehensive decision coordination for the creation of autonomous network management decisions. It does not offer a mechanism. For these reasons, the communication network of artificial intelligence agents with different areas of expertise They can independently evaluate the data obtained within their own areas of responsibility, The decision proposals generated by the agents are dynamic within a common decision-making process. This analysis shows that the priorities between different optimization goals are determined in the network's real-time. It can be managed according to the situation, and as a result, the communication network becomes more reliable and flexible. 25 A new system and method is needed that allows it to be managed in an explainable and autonomous way. It is heard. Figures that will help understand the invention. Figure 1 shows the general architecture of the system that is the subject of the invention. 30 Explanation of Part References 1. Network Data Collection Module 2. Data Preprocessing and Distribution Module 3. Artificial Intelligence Agents 35 4. Agent Coordination Module 5. Dynamic Decision Reconciliation Module 6. Conflict Resolution Module 7. Autonomous Network Decision Module 8. Network Management and Control Module 9. Performance Monitoring and Feedback Module 5 10. Learning and Model Updating Module Detailed Description of the Invention The invention enables the simultaneous achievement of 10 different optimization goals in communication networks. a multi-agent AI-based autonomous network management system that enables evaluation and relates to the method. The system uses operational data obtained from the communication infrastructure. data collection, processing, and analysis by artificial intelligence agents with different areas of expertise. the creation of decision proposals within a common decision-making mechanism based on the evaluation and application of the final decision to the network components 15 He is working. During system operation, the traffic density, network load, and service quality of the communication network are monitored. indicators such as energy consumption, user mobility, security events, connectivity status, and similar operational data are collected from different network components by the Network Data Collection Module (1) 20 This data is obtained. The collected data is analyzed within the Data Preprocessing and Distribution Module (2). By doing this, missing or erroneous records are removed, data integrity is ensured, and shared data is maintained. It is converted into a specific format and then transmitted to the relevant artificial intelligence agents. The processed data is evaluated simultaneously by Artificial Intelligence Agents (3). 25 Within the scope of the invention, each artificial intelligence agent is structured to specialize in a particular area. and analyze the same dataset independently in line with its own optimization goal. For example, one artificial intelligence agent might evaluate energy efficiency while another artificial intelligence agent does the same. One AI agent assesses service quality, another AI agent assesses security status, and yet another AI agent assesses... The intelligence agent, on the other hand, can analyze traffic density or resource usage. Thus, every 30 The artificial intelligence agent generates independent decision recommendations within its area of ​​expertise. Decision recommendations independently generated by artificial intelligence agents are processed by the Agent Coordination Module. (4) is transferred to the joint decision-making process managed by. This module transfers data between agents. It regulates the sharing, manages the decision-making sequence, and, when necessary, specific 35 enabling the agents to be reactivated and all agents to work in a coordinated manner. 6 This makes its operation possible. Thus, the system is not dependent on a single artificial intelligence model. It allows for the joint evaluation of different areas of expertise without any limitations. Decision proposals obtained at the end of the coordination process Dynamic Decision Reconciliation Module (5) This module is evaluated jointly by 5 different artificial intelligence agents. suggestions regarding the real-time status of the communication network, system priorities, and defined optimizations. By analyzing the situation in line with their objectives, artificial intelligence agents reach a consensus. If it produces different proposals for the same situation, the Conflict Resolution Module (6) is activated. This module identifies inconsistencies between decision proposals. Performing prioritization processes and making the most appropriate decision from a system perspective. 10 It enables the creation of an alternative. Thus, optimizations that can conflict with each other. The goals are to achieve a balanced and consistent decision-making mechanism. The resulting collective decision is then used by the Autonomous Network Decision Module (7) for final network management. This decision is being transformed into a resolution. This resolution involves the reallocation of network resources, 15 changing traffic routing, implementing energy management policies, security enabling of measures, restructuring network segments or communication It may include other control operations related to the operation of the network. The decision made is Network Administration. and the communication network is implemented by applying it to the relevant network elements via the Control Module (8). It is managed autonomously. 20 Following the implementation of the decision, network performance is continuously monitored through Performance Monitoring and Feedback. The Feeding Module (9) monitors and ensures that the system meets the defined performance targets. Whether it provides or not is analyzed. The feedback information obtained is used in Learning and The decision of artificial intelligence agents is evaluated by the Model Update Module (10) 25 Improving data delivery performance, updating model parameters, and the system continuous improvement to adapt to changing network conditions This is being done. Updated models are being reintegrated into the system, resulting in more efficient use of technology. an agent-based artificial intelligence-driven autonomous network management system that is learning, self-improving, and It works in a dynamic structure that can produce more accurate decisions over time. 30 is provided. This invention enables artificial intelligence agents with different areas of expertise to work in a coordinated manner. its work, effective management of decision differences, real-time communication network the creation of shared decisions that adapt to the situation and the 35% support for human intervention 7 by reducing the need, communication networks become more reliable, flexible, scalable and autonomous. This makes it possible to manage it in this way. 10

Claims

8 REQUESTS 1. Coordination of artificial intelligence agents with different areas of expertise in communication networks. Multi-Agent AI-Based System that performs autonomous network management through its work It is an Autonomous Network Management System; its features include:  Network traffic load, network status, resource utilization rates, energy consumption, 5 Network Data collects operational data such as service quality, security incidents, and similar data. Addition Module (1),  Pre-processing the collected data, ensuring data integrity, and relevant artificial intelligence. Data Preprocessing and Distribution Module (2), which distributes to its agents,  Analyze processed data independently according to different optimization goals. 10 Artificial Intelligence Agents (3) that generate decision recommendations,  managing the operational processes of artificial intelligence agents, regulating data sharing, and Agent Coordination Module (4), which ensures coordination between agents,  Real-time communication of decision recommendations generated by artificial intelligence agents through the communication network taking into account the situation, system priorities and operator policies, together 15 Dynamic Decision Consensus Module (5) which evaluates  Analyze the inconsistencies between decision recommendations obtained from different artificial intelligence agents. Conflict Resolution Module (6), which enables the formation of the appropriate decision,  Autonomous Network that transforms the collectively formed decision into an implementable network management decision Decision Module (7), 20  Traffic management and resource management by applying the decisions made to communication network components. allocation, network configuration, energy management, implementation of security policies or A network that performs at least one of the network segment restructuring operations. Management and Control Module (8),  Continuously monitor network performance, resource utilization, and quality of service parameters. 25 The Performance Monitoring and Feedback Module (9),  Using performance feedback to enable AI agents to make decisions Learning and Model Update Module (10) which updates its processes It includes.

2. It is an autonomous network management system according to Claim 1, and its feature is Dynamic Decision Reconciliation. The decision recommendations generated by the module (5) different Artificial Intelligence Agents (3) the current status of the communication network, system priorities, and operator-defined parameters. It is about reaching a common decision by jointly evaluating policy rules. 35 9 3. An autonomous network management system according to claim 1 or 2, with the feature of; Conflict Resolution. The conflicting decision generated by the module (6) different Artificial Intelligence Agents (3) analyzing the suggestions, prioritizing them, and making the final decision. It is a verification.

4. An autonomous network management system according to any of claims 1-3, and its feature is; Artificial Intelligence 5 (3) their agents' energy management, service quality optimization, security management, traffic management, resource optimization, user mobility analysis, or at least one of these. It consists of specialized artificial intelligence agents customized for this purpose.

5. An autonomous network management system according to any of the requirements 1-4, and its characteristic is: Network performance, service 10 of the Performance Monitoring and Feedback Module (9) quality, resource utilization rates, and feedback obtained from system outputs transmitting the data to the Learning and Model Update Module (10) and the said The module uses this data to analyze the decision-making processes of Artificial Intelligence Agents (3). This is an update. 20 30 35

Images