Joint Resource Management For Heterogeneous Converged Networks

In recent years,with the increasing complexity of the network scale,the types of services carried by the network are increasing day by day,and the quality of service and bandwidth resources demanded by different services show significant differences.On account of its large-scale and multi-access,a single-architecture network can no longer meet the network quality of service requirements,and heterogeneous network convergence of different types of access network resources is an inevitable trend in the development of next-generation networks.Due to the heterogeneity of access networks and the uncertainty of service requirements,traditional single network resource management solutions cannot be directly applied in heterogeneous environments.The urgent problem to be solved in heterogeneous converged network research is to centrally control the resources of different access networks and establish a unified heterogeneous network resource management system to ensure the transmission performance of the network.This thesis implements an intelligent resource control strategy for heterogeneous converged networks oriented to SDN network architecture with the separation of forwarding plane and control plane for multi-type access network convergence scenarios.The research is conducted from two aspects: heterogeneous network access and network bandwidth allocation.To solve the problem of multi-access network resource cooperative management,this thesis proposes a deep reinforcement learning heterogeneous network access selection strategy based on neuro-evolution strategy.Based on the characteristics of users,services,and networks,the system collaboratively manages and controls different access network resources,selects access networks with optimized performance,improves overall network service quality,improves network throughput and service delivery rate,and minimizes network switching costs.Additionally,to address the rational allocation of limited bandwidth resources in multi-service demand scenarios,this thesis proposes a network bandwidth resource allocation strategy based on the asynchronous multi-threaded A3 C algorithm.It provides flexible and adaptive network bandwidth resource management for multiple services in heterogeneous network architecture to maximize network performance and service utility.Compared with traditional algorithms,the bandwidth allocation strategy proposed in this thesis has certain advantages in terms of important service arrival rate,network throughput,and computation time.The characteristics of heterogeneous communication networks and the application of control and forwarding separation of SDN technology are studied.A heterogeneous converged network system platform based on SDN is built,and the heterogeneous network resource control algorithm is deployed and validated to improve the deployment ability and applicability of the heterogeneous network resource control algorithm.