Analysis of Autonomous Opportunistic Spectrum Access Strategies for Cognitive Radios: Benefits of Adaptive Strategies

We investigate autonomous opportunistic spectrum access (OSA) strategies for a distributed CR network, where two or more distributed CRs sense the channels sequentially in some sensing order until they find a free channel to transmit on, if one exists. First, we evaluate the performance of a traditional OSA strategy in terms of probability of success (probability that a given cognitive radio finds a channel free), mean time of success (if successful, mean time to find a channel) and throughput achievable by a CR. We find that the performance of the traditional OSA strategy is limited by the collisions among the distributed uncoordinated CRs. Then to reduce the likelihood of collisions among the distributed CRs we propose adaptive OSA strategies. We show that when the number of CRs is less than or equal to the number of potential channels the proposed adaptive OSA strategies enable the CRs to arrive at collision-free channel sensing orders. We also show that a γ-persistent strategy, in which a CR maintains information regarding past successes (and failures) of accessing the channel, converges quickly to collision-free sensing orders. We compare such a strategy against a randomize after every collision strategy, in which a new channel sensing order is randomly selected whenever a CR experiences a collision. Moreover, the implementation of γ-persistent strategy is not complex since it requires only two binary flags for maintaining state information at an individual CR.