The Research On Load Balancing Algorithm For Heterogeneous Wireless Networks Based On Computational Intelligence

The new generation of wireless communication system is a heterogeneous network which formed by the overlapping multiple wireless networks. Different networks have different wireless access technologies access technologies, and vary in terms of coverage, network architecture, transmission rate, mobile support etc. With the rapid growth of business, load imbalance occures between networks, causing worse quality of service, the declining rate of the whole wireless resource. So, How to allocate the limited source has become a hot and important research topic. The network load balancing is the key technology of radio resource management, can improve radio resource utilization based on guaranteeing the user quality of service,and the focus of this research.This paper summarizes the four most common heterogeneous network architectures and radio resource management technology, and analyzes the load balancing technology. On this basis, load balancing algorithms based on admission control and forced handover are proposed.Based on admission control, a load balancing algorithm based on fuzzy neural network is proposed, it concludes difference prediction, predecision, fuzzy neural network and access decision modules. In the predicision module, The algorithm sets different admission mechanismes for different user’s moving speed, and sets different admission processes for real-time and non real-time services. In fuzzy neural network module, the algorithm considers the user requirements and network load, taking received signal strength, bandwidth, network load and cost as the decision factors. The simulation result shows that the algorithm can effectively reduce the packet loss rate and load difference between networks, achieve the system load balancing.Based on forced handover, a load balancing algorithm based on utility function and fuzzy logic is proposed, which can handover some users from the overload network to the light load network. The algorithm defines different utility functions for real-time and non real time on access network selection. The simulation result shows that the algorithm can reduce the page response time, improve costomer satisfaction, achieve load balance effectively, improve the whole system wireless resource utilization.