An Algorithm of Regression Prediction for RSS Based on SVM with Granulating Information^*

This paper presents an algorithm of regression prediction for RSS (Received Signal Strength) based on SVM (Support Vector Machine) with granulating information, which can be used in selecting access network in heterogeneous wireless network environment. We normalized the RSS to increase accuracy of prediction at first, then implement the regression prediction for granulating RSS using SVM. Using this method, we can predict RSS of mobile terminal in the next 2 time point, so that the mobile terminal can start handover to access another network in advance and reduce the handoff latency. Simulation results show that the algorithm can predict RSS with better accuracy.