Research On Markov Decision Model Based Network Selection Algorithm For Heterogeneous Wireless Networks

In the future wireless communication networks,a variety of access technologies work together,to achieve heterogeneous network convergence,so that mobile users in any heterogeneous environment can make network selection anytime and anywhere,to meet the individual quality of service(Qo S)requirement of users while improving the overall heterogeneous network resource utilization.As an important research content in heterogeneous network radio resource management,the network selection system decides the best access network by considering the attribute parameters of the wireless network and the business requirements of different users to ensure the continuity of service quality and network access,having a certain research significance.The existing research results have solved the problem of network selection from different perspectives,but there are still many problems.First,a variety of wireless access technologies coexist in heterogeneous wireless networks,and it is difficult to manage them uniformly.Secondly,if only the current time network state to carry out network selection,it is difficult from a long-term perspective to optimize the entire heterogeneous wireless network system.Based on in-depth research on the key technologies of heterogeneous wireless networks,this thesis first envisages the future heterogeneous network architecture,introduces software defined network(SDN)and network function virtualization(NFV)technology to realize the flexibility of the network control and dynamic deployment of network functions,and then proposes a Markov decision model based network selection algorithm for heterogeneous wireless networks,by predicting the network status at the next time,the best access network can be selected for different service types users in the long run to meet the individual Qo S requirements of users.The main work of this thesis is as follows:1.A future heterogeneous wireless network architecture based on SDN/NFV is envisaged in this thesis,integrating the SDN controller module and the network function virtualization module in the control layer to shield the difference of various wireless access networks to realize unified management and control.Through data and control separating ideas of SDN,hardware and software decoupling ideas of NFV to achieve network flexibility,programmability and network virtualization capabilities,we can reduce the complexity of network management operations and promote the convergence of heterogeneous wireless networks.2.A Markov decision model based network selection algorithm for heterogeneous wireless networks is proposed in this thesis.Considering individual Qo S requirements of different service types,the Markov decision processes(MDP)model of network selection is constructed for real-time service users and non real-time service users respectively.Through the effective prediction of the next time network state to get the action after the return function,and use the analytic hierarchy process to determine the state attribute weight.In the long term,the objective function of the function sequence is used to measure the long-term returns in the future,and then iterate over the expected return function to provide the user with the best network selection strategy.3.Heterogeneous wireless network environment is built by Matlab simulation software in this thesis,and the algorithm of network selection is simulated and verified.The simulation results show that the algorithm can choose the best strategy for different service types,reduce the blocking rate of all users,and improve the resource utilization rate of heterogeneous wireless networks in the long run.