Cooperative spectrum sensing with multi-bits local sensing decisions in cognitive radio context

There are two important constraints for cooperative spectrum sensing in cognitive radio (CR) context. Firstly, as the CR system can only tolerate low transmitting overhead, the local sensing data must be compressed before transmitting. Secondly, many sophisticated data fusion techniques cannot be used in CR system because of the lack of the signal's prior knowledge. In this paper, we proposed a novel cooperative spectrum sensing scheme which adapts to different overhead tolerance and does not need any prior knowledge. This scheme consists of two main parts: the quantization schemes and the data fusion rule. Two different quantization schemes, according to whether or not the distribution functions of test statistic are known, are proposed. Correspondingly, the optimal data fusion rule of multi-bits decisions is derived. Furthermore, to make the optimal fusion rule more practical, an iterative scheme, which does not need any prior knowledge of the signal, is proposed to estimate the likelihood ratio of local decisions. Simulation results show that our scheme achieves better performance than the schemes with "OR" and "AND" combinations. Furthermore, it is also shown that the proposed scheme could achieve the theoretically optimal performance by only two or three bits quantization.