Spectrum Sensing Algorithm Based on Minimum Description Length Criterion for Sparse Multiband Signal

Spectrum sensing issue for sparse multiband signals is considered in this paper. Firstly, undersampling at sub-Nyquist rate are used to construct correlation matrix, and eigenvalue decomposition is carried out. Then, the method uses Minimum Description Length (MDL) criterion to distinguish between signal subspace and noise subspace and finds the spectral support. A spectrum sensing algorithm for sparse multiband signal is presented. Traditional methods distinguish between signal subspace and noise subspace by setting threshold, and make mistake when the threshold is wrong. Adaptability of this algorithm is better because there is no need to set threshold. The simulation results verify the validity of this algorithm.