Characteristic Utility Function And Comprehensive Weight Based Heterogeneous Network Selection Algorithm

The next generation of wireless communication system will integrate different wireless networks to provide a seamless access and diversified service for users. The basic feature of heterogeneous wireless system is that the system can select the best suitable access network for users to provide reliable guarantee of QoS based on the current network status and service types.Due to the heterogeneity of the future wireless communication system, the denotation method of resources allocated to uses to satisfy QoS requirements of services is very different in each different wireless networks. The diversities make it difficult to transfer the QoS requirements between different wireless networks seamlessly. Therefore, the equivalent spectral bandwidth concept is proposed in this thesis to map the different denotation of resources occupied by users in LTE and WLAN networks into spectral bandwidth. The equivalent spectral bandwidth occupied by each user in each network can clearly reflect the QoS requirements of their service. Additionally, the design of network selection algorithm relies on the comparison of the system residual resources between each wireless network, so the uniform denotation of resources is also important for the calculation of the system residual resources. Therefore, the equivalent spectral bandwidth, as a unified denotation of resources occupied by each users, simplifies the resource expression and the residual network capacity calculation in the heterogeneous wireless system.According to the equivalent spectral bandwidth model of LTE/WLAN heterogeneous system, a network selection algorithm based on characteristic utility function and comprehensive weight is designed in this thesis. On the basis of MADM, the proposed algorithm chooses suitable utility function to standardize each attribute according to its feature and its relationship with user satisfaction, and introduces single-objective optimization model to integrate subjective weight and objective weight to make a comprehensive weight which not only meets users’ preferences but also considers the system status simultaneously. The simulation results show that this network selection algorithm can achieve better load balancing performance, and reduce the vertical handoff frequency.