Performance analysis of a slotted multi-channel MAC protocols for cognitive radio networks

Cognitive radio has emerged as one of effective methods to enhance utilization of radio spectrum. Main principle of cognitive radio is that the secondary users (SUs) are allowed to use the spectrum not used by primary users (PUs) without interfering PU's transmission. In this paper, we consider network system where PUs use channels under super-slot time structure and SUs content to access channels during unused super-slot so as not to interfere PU's transmission. For contention resolution, super-slot is divided into slots of fixed size which are used as backoff unit for backoff algorithm. For contending SUs, our proposed MAC protocol operates by the following modified CSMA/CA with binary backoff algorithm: Each SU uniformly chooses a backoff counter from the current backoff window. The backoff counter indicates the number of slots that the station has to wait before the transmission. At the beginning of super-slot, SU senses pre-determined number of channels. If there are no idle channels, then the backoff counter of the SU is frozen during a current super-slot duration. If there are idle channels, then the SU decrements its backoff counter by one per each time slot as long as time slot is idle. When the backoff counter of the SU reaches zero, the SU transmits the packets as many of fixed size as idle channels accommodate in the current super-slot. During backoff procedure of the SU, if other SUs occupy the remaining idle channels, backoff counter of the SU is frozen during the remaining part of super-slot and is reactivated when at least one channel is sensed idle at the beginning of the super-slot. We construct the three-dimensional discrete time Markov chain (DTMC) to investigate the performance of the proposed multi-channel MAC protocol and we obtain stationary probability vector of DTMC by censored Markov chain's method. Then, using the stationary probability vector of DTMC, we obtain the head of line (HoL) packet delay distribution and the normalized throughput of SUs.