A Novel Spectrum Handoff Management Scheme based on SVM in Cognitive Radio Networks

Cognitive radio networks bring spectrum access opportunities to SUs(Second Users) when primary users would not use this spectrum so as to enhance the resource utilization. The SUs have to change spectrum channel no matter when and where the PU(Primary User) appears. Much research has been done on spectrum switching prediction design, without considering the spectrum handoff caused by the randomly moving of SUs and PUs. This paper proposes a novel spectrum handoff management scheme based on Support Vector Machines (SVM). To get a better spectrum sensing result, a cooperative spectrum sensing mechanism is used. The SVM model is introduced to predict the handoff point, so that the SU can prepare to handoff before the channel is occupied. The simulations results show that the algorithm could forecast the handoff point and obviously reduce the handoff time.