Research On Heterogeneous Wireless Network Load Balancing Technology Based On Time Series Prediction

The next generation wireless network will be a heterogeneous network with 2G, 3G, 4G cellular wireless network and WiMax, WLAN wide use in the world. Load balancing is one of the key technologies to realize the resource sharing in heterogeneous wireless networks. It can improve the utilization of network resources and reduce network congestion.Heterogeneous wireless network vertica hand off pay a crucial role in load balance. Through analysis SAW algorithm, Obtaining the correlation coefficient between load and call arrival rate, delay time, regression equation. We know shortcomings of the SAW algorithm in the presence of load delay and the low utilization rate of network resources. In this paper, we improve the SAW algorithm for these two shortcomings.Firstly, to overcome the load lag of SAW algorithm, base on the SAW algorithm, This paper studies the TSAW algorithm based on time series forecasting to improve SAW. Whith the use of call arrival rate in the time series forecasting, guide the business switch and bandwidth transfer in advance, we have overcome the load lag character of SAW algorithm.Second, aiming at the shortcoming of low resource utilization in SAW algorithm, based on the TSAW algorithm using the time series forecasting network call arrival rate value, to determine the execution of business bandwidth increase or reduction strategy to form the improved TBSAW algorithm. In addition, in order to make more effective TBSAW algorithm for network load balancing, designed the power-exponential function transfer call arrival rate, use predict call arrival rate as a base, predict call rate of variance as the index. That the improved algorithm when load is hight business bandwidth is less, when load is low business bandwidth is more, call arrival rate variance as the index, effectively restrain the load fluctuation.In this paper, SAW, TSAW, TBSAW algorithm design three kinds of simulation methods: First, load timing simulation diagram, clear observed TBSAW, TSAW algorithm contrast SAW appear cut peak, clipping, advance, smooth load characteristic. Second, in order to overcome the accident of timing simulation, time-varying, use Eviews software to make descriptive statistics, use statistical methods to do statistical analysis. Third, call blocking probability, analysis of three algorithm. Three methods of simulation analysis show that SAW, TSAW, TBSAW load characteristic turn better. Reflects the algorithm based on time series forecasting data, and base prediction bandwidth adjustment algorithm they all improve laod balancing features.