Spectrum detection of cognitive radio based on blind signal separation

Cognitive radio (CR) is viewed as a novel approach for improving the utilization of radio electromagnetic spectrum resource, and spectrum detection is an important step of CR. As the traditional methods are detecting every channel orderly, and it does not satisfy the real-time needs of CR, a detection algorithm based on blind signal separation is proposed in this paper. Through the sensors, the mixed observation signals are obtained, and they are decorrelated and denoised by signal whitening and wavelet denoising. All the primary signals in the detection channels with uncertain sequence are obtained at one time based on blind signal separation, and by making FFT of them, the signals' frequencies and channels' states could be easily estimated. Simulation shows our algorithm which is based on a real-time, could restructure the primary signals, and the estimation of channels' states is accurate.