An Iterative Threshold Selection Algorithm for Cooperative Sensing in a Cognitive Radio Network

In this paper, we propose a novel threshold selection scheme for OR-rule based spectrum sensing with energy detection. Unlike previous works on OR-rule cooperative sensing where a common threshold is used for all secondary users (SU) under the assumption that they have identical average SNR and noise power, the new threshold selection scheme is studied under the more general framework of possibly different average SNR and different noise power level at each SU. This iterative threshold selection scheme significantly outperforms the common-threshold approach in terms of the total probability of erroneous decisions, when the noise power at each SU differs by even just a few dBs (which is most likely the case for any practical radio system). On the other hand, even when all the SUs have identical average noise power and SNR, the new scheme is able to deliver similar performance compared to schemes using a common detection threshold. This iterative scheme is also shown to have fairly low implementation complexity.