Statistical matched filter based robust spectrum sensing in noise uncertainty environment

Spectrum sensing enables cognitive radio to obtain radio environment knowledge. Most of spectrum sensing encounters low signal to noise ratio (SNR) signal detection problem. However, non-ideal noise which is modeled as noise uncertainty degrades the sensing performance, and makes it unsatisfied for the sensing requirement. To deal with the problem, this paper proposes a new method based on the ratio of maximum-to-mean absolute value of matched filter output. The test statistic proposed is robust to overcome the noise uncertainty. This paper also finds the threshold of the method for a given probability of false alarm, and gives theoretical analysis of probability of detection. Simulation results show the proposed method could achieve better performance than conventional spectrum sensing methods in noise uncertainty environment.