Research on Mixed Matrix Estimation Algorithm Based on Improved Sparse Representation Model in Underdetermined Blind Source Separation System

The estimation accuracy of the mixed matrix is very important to performance underdetermined blind source separation (UBSS) system. To improve matrix, sparsity signal required. novel fractional domain time–frequency plane obtained by rotating after short-time Fourier transform. This represents fine characteristics in time and frequency domain. rotation angle determined global searching for minimum L1 norm make sufficiently sparse. points do not need single point detection, reducing calculation amount original algorithm, insensitivity noise improves robustness algorithm environment. simulation results show that are improved. Compared with existing algorithms, proposed method effective.