Optimization of cooperative spectrum sensing under noise uncertainty

Noise uncertainty, which is unavoidable in practice, can severely limit the detection performance of cooperative spectrum sensing. Derived from the conventional hard combination scheme, the improved one-out-of-N rule is presently the most effective algorithm to reduce the influence of noise uncertainty. However, in this algorithm the abandonment of local test statistic when it falls in the uncertainty region may lose some useful information. In this paper, first we analyze the optimality of the improved one-out-of-N rule and indicate that there should be an optimal number of cooperative users that minimizes the total error rate. Then, a new weighted hard combination scheme which makes full use of the uncertainty region is further proposed. The optimizations of weight coefficient and user number in this new scheme are also investigated. Numerical results show that the minimal total error rate and its corresponding user number of this scheme are both lower than those of the improved one-out-of-N rule.