A Novel Spectrum-Sensing Technique in Cognitive Radio Based on Singular Value Decomposition

In order to solve the problem that detection probability based on energy detector in cognitive radio is low under low Signal-to-Noise Ratio (SNR) conditions, this paper proposes a novel spectrum-sensing method for cognitive radio based on singular value decomposition (SVD). First, the received signal is represented in Hankel matrix, which is divided into signal space and noise space by using SVD. By ignoring the smaller singular values and then reconstructing the matrix, the SNR of received signal is increased. Then the power of the output signal of SVD system is normalized by noise power, and thus the reduction of noise power amounts to the improvement of user signal power. Finally, energy-detection is employed to the normalized signal, through which the detection probability is improved. Theoretical analyses and computer simulation results both shows that the detection probability of the proposed spectrum-sensing technique based on singular value decomposition is better than traditional energy-detection under the same circumstances. Besides, to achieve the same detection probability, the proposed method requires lower SNR.