Cognitive MAC Protocol with Transmission Tax: Probabilistic Analysis and Performance Improvements

We investigate the performance of a cognitive personal area network (CPAN) in which spectrum sensing is linked to packet transmissions. Efficient CPAN operation may be achieved if each data transmission is taxed by requiring the transmitting node to participate in cooperative sensing for a prescribed time period. In this approach, each node is allowed to transmit a single packet in one transmission cycle, but must then ‘pay’ for it by spectrum sensing, which ensures fairness with respect to transmission but also distributes the sensing burden to all nodes. We describe a probabilistic model of the integrated system and evaluate its performance with respect to packet transmissions and spectrum sensing. We discuss two modifications that involve centralized and distributed selection of the channels to be sensed. We also propose an adaptive algorithm to determine the tax coefficient and show that it offers superior data transmission performance whilst not affecting the sensing accuracy.