User numbers estimation for underlay Cognitive Radio

This paper is proposed a novel method of estimating the number of users for underlay Cognitive Radio (CR) to solve the difficulty of estimating the user numbers with time-frequency overlapped signals. The received signals are cut and decomposed with Singular Value Decomposition (SVD), firstly. And then, the period of the signal components can be estimated by detecting the peak of the Singular Value Spectrum (SVR) and finding the corresponding length of cutting numbers. Finally, the number of users is estimated by using the correlation coefficient contribution rate in the SVR spectrum. Simulation shows that the proposed method has better estimation performance.