Neural networks for adaptive vertical handover decision

Abstract

This paper focuses on mobility management between heterogeneous radio access networks (RANs). A fuzzy multi- criteria vertical handover algorithm enhancing the handover performance is proposed. This algorithm is based on fuzzy logic control. The fuzzy logic controller (FLC) takes into account multiple relevant criteria and rules based on prior knowledge of the network. A multi-layer perceptron (MLP) neural network is trained in order to learn the relationship between the FLC parameters and throughputs on UMTS and WLAN. Then, MLP inversion is performed in order to obtain the optimal parameters of the membership functions starting from throughput objective values. The performances of our handover algorithm are evaluated and compared to a conventional vertical handover algorithm by means of simulations. The proposed handover algorithm improves the network performances.