Noncooperative and Cooperative Joining Strategies in Cognitive Radio Networks With Random Access

Abstract

A cognitive radio system with retrial possibility and an admission cost for second users (SUs) to join the retrial group is investigated in this paper. If the SU finds the primary user (PU) band unavailable, it must decide with a probability estimate to either enter a retrial group or give up its service and leave the system. SUs in the retrial group independently retry after an exponentially distributed random time until they successfully access the spectrum. When the PU arrives, the SU's service on the band is interrupted. This interrupted SU is then assumed to occupy the PU band immediately when the PU completes its service. First, the non-cooperative joining behavior of SUs that choose to maximize their benefit in a selfish distributed manner is investigated, and an inefficient Nash equilibrium is derived. Second, from the perspective of the social planner, the socially optimal joining strategy when SUs cooperate with each other is studied and the corresponding Nash equilibrium is derived exactly. Finally, the result that an individually optimal strategy in general does not yield the socially optimal is verified theoretically. Furthermore, to bridge the gap between the individually and socially optimal strategies, a novel strategy of imposing an admission fee on SUs to join the retrial group is proposed and investigated with the derivation of optimal value for the admission fee. The numerical analysis indicates that the proposed admission fee as an equilibrium strategy and the socially optimal strategy of SUs improve efficiency in utilization of the cognitive radio system.

Game-theoretic analysis of opportunistic spectrum sharing with imperfect sensing

“…The works of Jagannathan et al [10] and Wang and Li [11] are closely related to this paper. The differences between this paper and Jagannathan et al [10] are as follows.…”

“…Several studies in the literature [8][9][10][11] considered the decentralized behavior of SUs and adopted queueing-game approach to investigate the interactions between PUs and SUs in the CR networks. Li and Han [8] studied the discrete model in a more applicable way which obtained the threshold of queue length to characterize the optimal joining strategy of SUs.…”

“…All of these works [8][9][10] assumed that there is a queue in front of the PU band and the new arrivals of SUs will enter the queue according to a firstin-first-out (FIFO) discipline. Recently, Wang and Li [11] considered the strategic behavior of SUs with retrials but assumed that the interrupted SU did not leave the service area and it would get service immediately once the occupied PU completed service and left the band. It should be noted that, in all the aforementioned papers, spectrum sensing is assumed to be perfect.…”

“…(2) It did not consider the sensing failure problem. Compared to Wang and Li [11], this paper assumes that the interrupted SU will leave the service zone and go back to retrial orbit as a head SU in the retrial queue. Therefore, a new arriving SU has a chance to utilize the PU band directly if the PU band is idle upon arrival.…”

“…In the work of Wang and Li [11], they assumed that the interrupted SU would not leave the service area and it will get service immediately once the occupied PU completes his service. As a result, the new arrivals of SUs during this period (the waiting period of interrupted SU in the service area) have to enter the retrial orbit for later attempts.…”