Research On Load Balancing Based Intelligent Access Selection Algorithm In Heterogeneous Wireless Networks

The next-generation wireless communication systems will be the heterogeneouswireless networks which integrate a variety of radio access technologies to support theusers’ seamless mobility and provide high-speed wireless connectivity for mobile users. Inheterogeneous wireless networks, the joint radio resource management of wirelessnetworks can reduce the unbalancing of network resource allocation and improve theutilization of radio resource. Therefore, it is of great significance to research on selectingthe appropriate access networks for users to improve users’ satisfaction, avoid networkcongestion and balance the load between networks.With the development of wireless communication technology, the heterogeneity ofsystems and support for multi-service make the uncertain and dynamic characteristics ofthe system load very prominent. The artificial intelligence technologies, such as fuzzy logicand neural network, have no need to establish accurate system model and can simplify thecontrol of uncertain and dynamic systems. So the artificial intelligence technologies have avery broad space for application and development in the radio resource management ofheterogeneous wireless networks.Aiming at working out the problem that fuzzy logic and neural network based accessselection algorithm didn’t consider the load state reasonably in heterogeneous wirelessnetworks, a RBF (Radial Basis Function) fuzzy neural network based access selectionalgorithm (RBF-FNN) is proposed. The algorithm takes the amount of the networkresources and users’ signal strength into consideration and adopts pre-decision andpretreatment process to reduce the algorithm overhead. The algorithm executes factorsreinforcement learning for the fuzzy neural network with the objective of the equalblocking probability of accessible networks to adapt for load state dynamically andachieved the intelligent access judgment. The simulation results show that the algorithmcan balance the load of heterogeneous wireless networks effectively and guarantee thequality of service, as well as decrease the blocking probability compared to the maximumload balance based algorithm (MLB algorithm).In order to reduce the algorithm complexity, a PSO (Particle SwarmOptimization)-fuzzy neuron based access selection algorithm is proposed. The algorithmtakes a fuzzy neuron as the main part of the access judgment controller, and combines withPSO algorithm with global optimization capability to set initial parameters value, so as to improve the precision of parameter learning. Simulations show that the performance ofPSO fuzzy neuron algorithm is similar to that of RBF-FNN algorithm, among which theaccess blocking probability, packet outage probability and the level of load balance areclose. The performances of proposed two algorithms are better than MLB algorithm. PSOfuzzy neuron algorithm can reduce algorithm complexity while ensure the performance andpracticality.The research work and results of this paper could provide references for the researchon load balancing based access selection algorithms in heterogeneous wireless networks.