Network Selection For Heterogeneous Wireless Networks Based On Reinforcement Learning

With the rapid development of mobile Internet and the diverse needs of users on the network, operators in order to improve the market competitiveness so that various wireless access technologies have been developed. The network system which is built up by different wireless technology is the heterogeneous wireless network system, and the coexistence of different wireless networks is the current situation of Internet development.A variety of terminal equipment in the market generally have multiple mode interface, they can choose different wireless access network to access in. Therefore, the network selection problem is one of the major issues which the integration of heterogeneous wireless networks is to be solved. The major service industries promote people-centered, and insist that customer is God, therefore this paper focus on the experience of the users to study the wireless network selection from the user's point of view. On the other hand, in order to avoid network congestion which can cause the decline of the quality of service, we charge the service to control it.This paper first introduces the development of mobile communication and the current situation of various wireless access technologies, have pointed out that the convergence of heterogeneous wireless is the inevitable developing trend, and have summarized the different current network selection method in heterogeneous wireless network. Then it introduces the influencing factors, attributes and quantitative methods of QoE in detail.Comparing QoS and QoE, we find that the importance of considering from the user's QoE.Finally this dissertation introduces the reinforcement learning model in psychology, and models the network selection by analyzing the problems. We propose an network selection algorithm for heterogeneous wireless network. The simulation results show that this method can effectively improve the experienced performance of the users, at the same time it can makes the network load balancing.