MAC protocol classification in a cognitive radio network

Abstract

Most media access control (MAC) protocols can be classified as contention based or controlled based according to their transmission mechanisms. To classify contention based or control based MAC protocols in an unknown primary network, we choose received power mean and variance as two features for support vector machines (SVMs) in a machine learning based algorithm. The data consisting of these two features are collected from two primary network models based on time division multiple access (control based) and slotted Aloha (contention based), respectively. In the training process, data along with their identification class labels (say, 1 denotes time division multiple access, and -1 stands for slotted Aloha) are used to train the SVMs. After training, contention or control based MAC protocols can be effectively determined by the trained SVMs embedded in a cognitive radio terminal of a secondary network.