Research On Intelligent Control Technology Of Fiber-Wireless Converged Network For Edge Computing

With the advent of the 5G data era,the number of access devices in the network has increased exponentially,and the flood of data has exploded,putting forward new requirements for ultra-low latency and high reliability of the network.The centralized core network of the traditional centralized processing mode can no longer support the development of new requirements.Edge computing based on the optical and wireless converged network is located at the edge of the network,close to the user side,and has the advantages of large-capacity and high-speed transmission of the optical network.,and has the flexibility advantages of wireless networks,and has become a new networking architecture to meet the new requirements of 5G.In 5G application scenarios,the diversification and differentiation of network user requirements pose new challenges to the flexibility and customization of network resource management and control.The traditional rigid network resource management method is difficult to adapt to edge scenarios,and the intelligent development represented by artificial intelligence technology provides a new idea for it.This paper focuses on the research on the intelligent management and control system of Fiber-Wireless Converged Networks for Edge Computing(FiWi-ECN)based on edge computing,and has obtained several innovative research results,including:First,in response to the problem that the current resource management method cannot adapt to the differentiated needs of services generated by 5G diverse scenarios,this paper studies the artificial intelligence-based FiWi-ECN network management and control system,and designs its network architecture,including user demand mapping model.,network resource intelligent configuration and resource scheduling policy guarantee mechanism,form a solution to meet network user service requests,reduce network abnormalities,and improve resource utilization.Second,in view of the problem of too rigid resource allocation in FiWi-ECN network,a resource allocation strategy generation mechanism based on deep reinforcement learning is studied,and a new resource allocation method for FiWi-ECN network is generated by combining various strategy actions,which realizes user Optimal adaptation of service requests and resource allocation strategies.Simulation results show that the proposed strategy model improves the network resource utilization by 32.1%.Third,for the problem that resource allocation strategies cannot be optimally adapted to sudden changes in the network environment,this paper studies anomaly detection methods based on deep learning,and proposes an unsupervised clustering and supervised learning method as a security mechanism for allocation strategies.The simulation results show that the accuracy of anomaly detection model can reach 94.36%and the anomaly detection time can be shortened by 0.6s under the information of tens of thousands of scale datasets.