Fuzzy Neural Network Based Vertical Handoff Decision Algorithm In Heterogeneous Networks

With the rapid development of wireless communication technology, coexistence andintegration of different access networks has become the inevitable trend of future wirelessnetworks. In the coexistence of a variety of techniques for heterogeneous wireless networkenvironment, as the basis for integration and coordination of heterogeneous wirelessnetworks, the research on vertical handoff has far-reaching significance to optimize loadmanagement of wireless networks and to provide users with reliable service.Most strategy and multi-attribute decision based vertical handoff algorithms need toestablish exact model for describing the handoff decision parameters, but not all thehandoff decision parameters can be exactly modeled by the algorithms. What’s more, thealgorithms have to modify the model in time which will cause the manual interventionprocess to adapt to the network changes. To some extent, the neural network and fuzzylogic based algorithms can overcome the above problems, and the algorithms are the betterchoice to realize the intelligent vertical handoff in heterogeneous wireless networks.This paper applies the fuzzy processing and self-learning ability of fuzzy neuralnetwork to vertical handoff decision of heterogeneous wireless networks. Aiming atworking out the problem that fuzzy neural network based vertical handoff decisionalgorithm didn’t consider the load state reasonably in heterogeneous wireless networks, aparticle swarm optimization(PSO)-FNN-based vertical handoff decision algorithm isproposed. When the user requests to handoff, the PSO-FNN algorithm takes networkbandwidth, the user received power as the important factors of the handover decision. Theproposed algorithm executes factors reinforcement learning for the FNN with the objectiveof the equal blocking probability and realizes the “passive load balancing” betweennetworks. Meanwhile, PSO algorithm is adopted to set initial parameters of FNN in orderto improve the precision of parameter learning. The simulation results show that thePSO-FNN algorithm can balance the load of heterogeneous wireless networks effectivelyand decrease the blocking probability as well as handoff call blocking probability comparedto sum-received signal strength (S-RSS) algorithm.On the basis of the above study, a flow diversion-based vertical handoff algorithm(FDVHA) which adopts FNN to divide flows is proposed to work out the problem that thetraditional hard load balancing algorithms can’t make use of the network resourcesefficiently and the existing soft load balancing algorithms have application limitations. FDVHA algorithm decides whether to execute vertical handoff according to whether thenetwork load is balanced or not. Firstly, the algorithm chooses a user who needs to handoffaccording to the corresponding strategy. Secondly, FNN is adopted to obtain the optimalflow-dividing ratio. At last, user’s traffic is divided into each network according theoptimal flow-dividing ratio to realize the “active load balancing” between networks.Simulation results show that, compared to traditional hard load balancing algorithm,FDVHA can balance the load of heterogeneous wireless networks effectively and theaccess blocking probability is decreased by2.50%as well as the handoff call blockingprobability is decreased by2.70%.The research work and results of this paper could provide references for the researchon fuzzy neural network based vertical handoff decision algorithms in heterogeneouswireless networks.