Cooperative spectrum sensing has been shown to yield a significantperformance improvement in cognitive radio networks. In this paper, we considerdistributed cooperative sensing (DCS) in which secondary users (SUs) exchangedata with one another instead of reporting to a common fusion center. In mostexisting DCS algorithms, the SUs are grouped into disjoint cooperative groupsor coalitions, and within each coalition the local sensing data is exchanged.However, these schemes do not account for the possibility that an SU can beinvolved in multiple cooperative coalitions thus forming overlappingcoalitions. Here, we address this problem using novel techniques from a classof cooperative games, known as overlapping coalition formation games, and basedon the game model, we propose a distributed DCS algorithm in which the SUsself-organize into a desirable network structure with overlapping coalitions.Simulation results show that the proposed overlapping algorithm yieldssignificant performance improvements, decreasing the total error probability upto 25% in the Q_m+Q_f criterion, the missed detection probability up to 20% inthe Q_m/Q_f criterion, the overhead up to 80%, and the total report number upto 10%, compared with the state-of-the-art non-overlapping algorithm.
展开▼
