AUTONOMOUS TELECOMMUNICATIONS NETWORK MANAGEMENT SYSTEM

Abstract

This invention relates to an autonomous network management system (1) that includes rollback and feedback mechanisms, enabling network interventions to be simulated on a digital twin before being implemented on the live network by analyzing real-time and historical data of the telecommunication network, evaluating the interventions according to risk and policy conditions, and carrying out appropriate interventions in a controlled manner based on artificial intelligence.

Description

1 TARIFF AUTONOMOUS TELECOMMUNICATIONS NETWORK MANAGEMENT SYSTEM Technical Area The invention is a system for managing and operating telecommunication networks. It is related to. Previous Technique Monitoring the status of network components in existing telecommunication networks, monitoring performance indicators, alarm and event logs evaluation and interventions for the malfunctions that occur Various network management systems are used for this purpose. Telemetry, alarms, logs, and performance data received from network components in systems 15 By analyzing the data, anomalies in the network can be detected. However, The problem was identified through monitoring network data in current applications. The intervention is often carried out by an operator or There are already defined rules in place. Especially on a large scale. and in telecommunication networks consisting of interdependent network components, a 20 Changes made to one component affect other components and services. It becomes difficult to accurately assess its impact beforehand. This Therefore, a new system is needed to address all the existing shortcomings. Patent number US20200196155A1, which is included in the prior art, 25 the document describes data collected from a 5G (fifth generation) telecommunications network. using machine learning-based closed-loop network management A solution for its implementation is described. This solution involves: Data obtained from the network is analyzed, and depending on the analysis results, the network... Procedures related to its management are being carried out. However, the current 30 In solutions, interventions to be implemented on the network are transferred to the live network environment without 2 first, it should be simulated on a digital twin representing the current state of the network, depending on the risk score and policy conditions of the interventions in question evaluation and controlled execution based on the evaluation results. There is no integrated structure for this purpose. In addition, the aforementioned document states that, Monitoring the results of the interventions performed, the expected 5% of the intervention assessing whether it has produced an effect and, if necessary, taking precautions. closed by activating a defined rollback mechanism There is also no structure in place to ensure the safe continuation of the cycle. Brief Description of the Invention 10 The purpose of this invention is to provide a real-time network status update for the telecommunications network. creating a digital twin that represents the data and on the network The digital interventions to be carried out will be reviewed before being implemented on the live network. The goal is to provide an AI-based network management system that simulates a twin. 15 Detailed Description of the Invention The "Autonomous Telecommunication Network" was created to achieve the purpose of this invention. The "Management System" is shown in the attached figure; this figure is 20. Figure 1. Schematic view of the system described in the invention. The parts shown in the figure are individually numbered, and the corresponding numbers correspond to these numbers. given below: 25 1. System 2. Electronic device 3. Database 4. Server 30 D. External server 3 A representation of the current status from real-time telecommunications network data. the creation of a digital twin, possible interventions on said digital twin simulation, AI-assisted analysis and intervention enabling the evaluation of proposals within the context of risk and policy conditions. system (1); 5 - reviewing intervention suggestions made by the user, Displaying information regarding approval or rejection processes and presenting intervention suggestions and procedure results to the user. at least one electronic device configured to provide an interface that enables (2), 10 - Network topology, network performance indicators, traffic information, alarms and events. logs, configuration information related to network functions, operation service inventory of the service support system (OSS) and business support system (BSS) and Configuration information, service inventory and configuration data, maintenance window logs, change history, past intervention results, digital 15 twin parameters, feature vectors, risk scores, policy information, record intervention logs, recovery plans, and operation results. at least one database configured to receive (3), - data exchange with electronic device (2), database (3) and external server (D) to perform, actual 20 from external server (D) to telecommunication network to collect time-based and historical data, to process the collected data, Creating a telecommunications digital twin, manipulating the digital twin. simulating scenarios, performing AI-based analysis, formulating intervention recommendations, analyzing intervention recommendations in terms of risk and policy. to evaluate within the scope of the conditions, to carry out appropriate interventions 25 to ensure, monitor network status after intervention and obtain at least structured to provide feedback from the results to the simulation. It includes a server (4). The system in question consists of (1) electronic devices (2), smartphones, tablet computers, 30 It is a device similar to a desktop computer or a portable computer. The one in question... 4 electronic device (2), any remote communication device in the known state of the art to establish a connection with the server (4) using the protocol and this established connection It is configured to exchange data with the server (4) via. In the preferred application of the invention, the electronic device (2), server (4) and data bus 5 to exchange data using the Internet It is being structured. In the system that is the subject of the invention, the database (3) in (1) is in communication with the server (4) and is configured to be managed by the server (4). The invention is preferred In the regulation made, the database (3) is the actual 10 related to the telecommunication network. time and historical data, network topology and configuration information, traffic and performance metrics, alarm and event logs, OSS / BSS service inventory, and configuration data, maintenance window logs, change history, anonymized usage intensity data, interventions carried out, intervention results, risk assessment results and feedback data 15 It is structured to keep records. The server (4) in the system (1) which is the subject of the invention, any remote communication to communicate with the electronic device (2) using the protocol and the established protocol Data exchange between electronic device (2) and external server (D) via communication 20 It is configured to perform. The server (4) is placed inside the database (3). making new data entries, deleting the registered data in the database (3) or Modifying the registered data in the database (3) and database (3) By updating the registered data in the database (3) It is configured to manage the server (4), live network topology, 25 Inter-node traffic metrics, alarm and event logs, performance counters, OSS / BSS service inventory and configuration data, maintenance window logs, including change history and anonymized usage intensity data, In order to represent the current network status of the digital twin and to update it... real-time data relating to the telecommunications network, message queue and 30 to receive from an external server (D) in the form of a data stream infrastructure The server (4) is structured into a common data schema of the received data. convert, align timestamps, complete missing data, outliers filtering values, performing normalization and feature extraction operations. It is configured to store the processed data in the database (3). It is structured to process the real-time and historical data. Server (4) 5 using a digital twin that represents the current state of the telecommunications network It is configured to create. Server (4), network topology, service to represent dependencies, capacity, traffic, operational status and customer impact It is structured as follows: Server (4), network elements on the digital twin. Modeling the relationships between them graphically, identifying causal relationships 10 and to evaluate hypothetical (what-if) scenarios regarding possible interventions. It is structured accordingly. The server (4) is planned to be implemented on the live network. to simulate the interventions primarily in two stages on a digital twin is configured. Server (4), in the first stage intervention normal operation the expected effect under these conditions, and in the second stage, the minimum that the intervention could create is 15 It is configured to assess adverse conditions and disruptive effects. Server (4), to evaluate the results of both simulations within the scope of safety thresholds and to prevent unsafe interventions from being implemented on the live network It is configured. Server (4), digital twin simulation results and real 20 to perform AI-based analysis on real-time network data It is structured. The server (4) detects anomalies, root causes within the scope of analysis. analysis, impact estimation, intervention recommendation formulation, and intervention prioritization. It is configured to perform the operations. The server (4) is used to perform the said analysis. time series analysis, statistical bias analysis, unsupervised learning, event Using methods such as correlation and graphical dependency analysis 25 It is structured as follows: Server (4), within the scope of artificial intelligence-based analysis. Isolation Forest, Autoencoder, XGBoost, Using models such as LightGBM and long short-term memory (LSTM). It is structured in such a way. The server (4) is reinforced for the purpose of intervention selection. learning (reinforcement learning), contextual multi-arm armband (contextual 30 (bandit), leveraging multi-objective optimization or constrained decision algorithms 6 It is structured accordingly. The server (4) is the target base in intervention selection. improvement of performance indicators (KPIs) and service level agreement (SLA) reducing the likelihood of violations and limiting the risk of interference It is structured to be evaluated together. Server (4), service interruption, Number of affected users, security risk, number of affected network elements, back 5 availability, past intervention success, regulatory compliance, and operational to calculate the risk score using parameters such as complexity The server (4) is configured with the risk score, risk score = w1 × SLA violation. Probability + w2 × KPI disruption effect + w3 × domain + w4 × difficulty of undoing + Calculating as w5 × policy non-compliance penalty + w6 × security impact, totaling 10. is configured. The server (4) sets the calculated risk score to the first threshold value (T1) and is configured to compare with the second threshold value (T2). The server (4), A risk score below the T1 value indicates a problem with the simulation on the digital twin. the results must meet safety thresholds, policy conditions must be fulfilled, and If the defined risk budget is not exceeded, the intervention will be automatically triggered. It is configured to execute. The server (4) has a risk score between T1 and T2. If this occurs, the intervention is structured to be submitted for human approval. If the server (4) has a risk score equal to or above the T2 value, to reject, postpone, or resend the intervention to a new simulation It is being configured. The server (4) will perform 20 actions within a specific time interval. to assess the total risk of interventions within the defined risk budget. It is configured accordingly. Server (4), traffic density, service criticality level, Risk depends on the maintenance window, security alarm level, and customer influence. It is configured to dynamically update its thresholds and risk budget. Server (4) has a predetermined 25 for each intervention to be implemented on the live network. to ensure there is a rollback plan and a rollback plan It is structured to prevent the implementation of interventions that are not currently in place. Server (4) approves interventions, target selection, command sequence, including dependency checking, pre-execution validation, and fallback flows It is configured to transform into controlled workflows. Server (4), software 30 Standardized Network (SDN) controllers, operational support systems (OSS), business support 7 systems (BSS), network management systems, configuration services, and robotic processes by communicating with automation (RPA) interfaces, routing parameters updating, increasing traffic capacity, re-services initiation, support log, configuration synchronization, field operation It is configured to enable the creation of requests and work orders. Server (4), 5 By performing conflict, dependency, and combined side effect analysis for multiple interventions. It is configured to determine the execution order. The server (4), intervention To record the result, return code, error information, and execution time, and to intervene. subsequently, delay, packet loss, error rate during a specific monitoring window, Service availability, customer impact, alarm status, traffic density, and resource 10 It is configured to monitor consumption. The server (4) displays the results obtained. expected improvement by comparing with results predicted by digital twin failure to occur, new alarm generated, increased customer impact, or service to identify instances of deterioration in performance and to intervene to change it. 15 to causally verify whether it originated from or not is being configured. Server (4), anomaly or causal deviation detected in that case, to activate the predetermined rollback plan is configured. The server (4) returns the results of successful and unsuccessful interventions. Using feed data, digital twin parameters, risk coefficients, Updating contextual thresholds, risk budget, and rollback triggers 20 and to transfer the updated results to subsequent decision-making processes. is structured. The server (4) learns from the results of the interventions and then proceeds. updating intervention decisions and closed-loop autonomous network management. It is structured to enable its implementation. The server (4) is artificial intelligence. Intervention proposals generated by [company name] are directly implemented on the live network. 25 to prevent its implementation and to assess the risks of the proposals in question through digital twin simulation. and to execute it in a controlled manner after it has undergone policy verification. It is being structured. 8 Industrial Application of the Invention The system that is the subject of the invention (1) shows the real-time status of the telecommunication network. using a digital twin representing the network and an AI-based analysis mechanism This ensures that interventions are carried out in a safe and controlled manner. 5 System (1) analyzes live network data and historical data to determine the current state of the network. It represents the situation on a digital twin, and potential interventions are being monitored on the live network. Simulating it on a digital twin before implementation and analyzing the simulation results It is evaluated within the scope of risk and policy conditions. If deemed appropriate... Interventions are carried out in a controlled manner, and the network obtained after execution is 10. The outcome of the intervention is evaluated by monitoring the data. As a result, the invention... Digital twin simulation, AI-based decision support, risk and policy. verification, controlled intervention, rollback, and feedback Closed-loop autonomous network management by using these mechanisms together. It is carrying out. 15 Around these fundamental concepts, the invention is called "Autonomous Telecommunications Network Management". It is possible to develop a wide variety of applications related to the System (1)”, and the invention This cannot be limited to the examples described here, but is primarily stated in the claims. It is like that.

Claims

9 REQUESTS 1. Represents the current status from real-time telecommunications network data. the creation of a digital twin, and possible modifications on said digital twin Simulating interventions, AI-assisted analysis 5 implementation and intervention proposals, risk and policy conditions enabling it to be evaluated within this scope; - information regarding transactions performed by the user to be displayed and intervention suggestions generated by the system, and 10. Providing an interface that allows the results of the process to be presented to the user. at least one electronic device configured to (2), - Network topology, network performance indicators, traffic information, alarms and events. logs, configuration information regarding network functions, OSS / BSS services inventory and configuration data, maintenance window logs, changes History, past intervention results, digital twin parameters, feature 15 vectors, risk scores, policy information, intervention records, rollback at least structured to record plans and transaction results a database (3) and to exchange data with electronic devices (2) and databases (3), external real-time and historical data from sources (D) regarding the telecommunication network 20 acquiring data, processing the acquired data, creating a digital twin, digital twin Simulating intervention scenarios, AI-based analysis. to implement, to formulate intervention proposals, to assess the risks of intervention proposals and to evaluate within the context of policy conditions, appropriate interventions to ensure its implementation, monitor network status after intervention and 25 At least one server configured to feed back the results obtained (4) a system characterized by its inclusion (1).

2. Real-time and historical data relating to the telecommunications network, network topology and configuration information, traffic and performance metrics, 30 alarm and event logs, OSS / BSS service inventory and configuration data, maintenance window logs, change history, anonymized usage intensity data, performed interventions, intervention outcomes, risk assessment results, and feedback. Database structured to keep feed data on record (3) A system like the one in Claim 1, characterized by (1). 5 3. Live network topology, inter-node traffic metrics, alarm and event logs, performance counters, OSS / BSS service inventory and configuration data, Maintenance window logs, change history, and anonymized usage. 10 related to the telecommunications network, including density data data from an external server in the form of a message queue and data flow infrastructure (D) is characterized by the server (4) configured to receive Request 1 or a system like in 2 (1).

4. Converting the collected data into a common data schema, including timestamps. 15 aligning, completing missing data, filtering outliers, to perform normalization and feature extraction operations on the processed material to store the data in the database (3) and to process real-time and historical using data to represent the current state of the telecommunications network 20 characterized by a server (4) configured to create a digital twin a system like any of the above-mentioned requests (1).

5. Network topology, service dependencies, capacity, traffic, operational status, and to represent customer influence between network elements on a digital twin modeling relationships graphically, identifying causal relationships, and 25 evaluating hypothetical scenarios regarding possible interventions, live network The interventions planned to be implemented on it will first be done on the digital twin. with the server (4) configured to simulate in two stages a system like any of the above characterized claims (1). 30 11 6. Initial impact of the intervention under normal operating conditions, In the second stage, the worst-case scenario and disruptive effect of the intervention are examined. to evaluate the effects, both simulation results against safety thresholds to evaluate within this scope and unsafe interventions on the live network Server (4) and 5 configured to prevent its implementation on it a system like any of the above characterized claims (1).

7. Through digital twin simulation results and real-time network data Performing AI-based analysis, identifying anomalies within the scope of the analysis. identification, root cause analysis, impact estimation, intervention recommendation formulation and to perform intervention prioritization procedures and the analysis in question for time series analysis, statistical bias analysis, unsupervised learning, using methods such as event correlation and graphical-based dependency analysis The above 15 is characterized by the server (4) configured to provide a system like any of the requests (1).

8. Isolation forest, autoencoder within the scope of AI-based analysis, XGBoost, LightGBM, and short-term and long-term memory models. contextual reinforcement learning for use and intervention selection 20 multi-arm armband, multi-objective optimization, or constrained decision server (4) configured to take advantage of algorithms a system like any of the above characterized claims (1).

9. Targeting key performance indicators in intervention selection to improve and reduce the likelihood of service level agreement breaches and assessing the need to limit the risk of intervention, service outage, number of affected users, security risk, affected network element number, retrievability, past intervention success, regulatory compliance and 30 Risk score using parameters such as operational complexity. 12 characterized by the server (4) configured to calculate a system like any of the above requests (1).

10. Risk score = w1 × Probability of SLA violation + w2 × Impact of KPI disruption + w3 × domain + w4 × rollback difficulty + w5 × policy non-compliance penalty + 5 Calculating the risk score as w6 × security impact, the calculated risk score is the first Comparing the risk score with the first threshold value (T1) and the second threshold value (T2) A T1 value below [a certain level] indicates that the simulation results on the digital twin are incorrect. meeting security thresholds, fulfilling policy conditions, and If the defined risk budget is not exceeded, the intervention will be automatic. The above is characterized by the server (4) which is configured to do so. a system like any of the requests (1).

11. If the risk score is between T1 and T2, intervention is necessary. to submit for approval, risk score must be equal to or above the T2 value 15 In that case, the intervention can be refused, postponed, or re-simulated. sending, interventions to be carried out within a specific time frame to assess total risk within the scope of the defined risk budget from the above requests characterized by the configured server (4) a system like any other (1). 20 12. A predetermined method must be used for each intervention to be implemented on the live network. to ensure a recovery plan is in place and for those for whom a recovery plan does not exist to prevent the implementation of interventions and to target approved interventions selection, command sequencing, dependency checking, pre-execution 25 controlled workflows including verification and feedback flows characterized by the server (4) configured to convert a system like any of the above requests (1).

13. Software-defined network controllers, operation support systems, business support network 30 management systems, configuration services, and robotic process automation 13 Updating routing parameters by communicating with interfaces, Increasing traffic capacity, restarting services, support registration, configuration synchronization, field operation request and work order structured to carry out interventions in the form of creating 5 in any of the above requests characterized by server (4) such a system (1).

14. Conflict, dependence, and combined side effect analysis for multiple interventions. determining the execution sequence, intervention result, return code, error to record the information and execution period and a specific 10 after the intervention latency, packet loss, error rate, service throughout a monitoring window availability, customer impact, alarm status, traffic density, and resources characterized by the server (4) configured to monitor consumption a system like any of the above requests (1).

15. Conflict, dependence, and combined side effect analysis for multiple interventions. determining the execution sequence, intervention result, return code, error to record the information and execution period and specific actions taken after the intervention. latency, packet loss, error rate, service throughout a monitoring window availability, customer impact, alarm status, traffic density, and resource 20 characterized by the server (4) configured to monitor consumption a system like any of the above requests (1).

16. Compare the obtained results with the results predicted by the digital twin. Comparing the two, the expected improvement did not occur, a new alarm was raised, 25 increased customer influence or deterioration in service performance to determine their status and whether the change is due to intervention causally confirming that it does not originate from, an anomaly or causal If a deviation is detected, the predetermined recall plan will be applied. 30 characterized by the server (4) configured to be put into operation a system like any of the above requests (1). 14 17. Successful and unsuccessful intervention outcomes as feedback data. using digital twin parameters, risk coefficients, contextual thresholds to update values, risk budget and buyback triggers and 5 to transfer the updated results to subsequent decision-making processes from the above requests characterized by the configured server (4) a system like any other (1).

18. Learning from the results of interventions to make decisions about future interventions. Updating and implementing closed-loop autonomous network management 10 to provide intervention recommendations generated by artificial intelligence to prevent it from being executed directly on the live network and the said recommendations from risk and policy validation through digital twin simulation server configured to execute in a controlled manner after passing through (4) like any of the above-mentioned claims characterized by 15 system (1). 25

Images