An MADM network selection approach for next generation heterogeneous networks

Next generation wireless networks will consist of different types of radio access technologies. Emerging diverse applications have attracted a lot of users. Therefore, having the best connectivity anytime anywhere has become an important issue. Users would like to have continuous high quality network connection even when they handover from one network type to the other. Each access technology has its own features and service quality. In this regard, when there is a handover, it is important to choose the most suitable network based on user needs. Different algorithms have been proposed for network selection. Multi Attribute Decision Making (MADM) methods are widely used due to their precision and low computational complexity. In this paper, usual normalization methods used in network selection and their drawbacks are presented and a new method, by taking user preferences and application needs into account, is proposed. Based on the proposed normalization method, a new decision process is suggested which is based on ELECTRE II algorithm. Our proposed method considers the dynamic nature of wireless networks and the uncertainty in the collected network performance metrics. Simulation results show that using the proposed handover decision process, the probability of ranking abnormality, which is a common drawback of MADM methods, is reduced significantly compared to other methods such as SAW, GRA and TOPSIS and the decision algorithm is more stable in terms of sensitivity to change in the amounts of decision criteria.