SVM-based Spectrum Handoff Scheme for Mobile Cognitive Radio Networks

Abstract

Few of the existing spectrum handofi algorithms take time-space-domain spectrum mobility into account for Cognitive Radio Networks (CRNs). In this paper, a novel spectrum handofi scheme based on the feature vector extraction of Support Vector Machine (SVM) is proposed. The new handofi scheme predicts the channel availability considering the activity of PUs and the mobility of SUs together. And Maximum Independent Set (MIS) theory is utilized to achieve fairness during the backup channel selection. Simulation results validate that the presented spectrum handofi algorithm obtains higher throughout fairness performances and lower collision rate than the traditional handofi schemes.