Abstract:

The underdetermined mixture matrix in blind source separation can be obtained by canonical decomposition of the threeorder tensor. In order to overcome the flaw of high computational complexity and long running time of existing canonical decomposition algorithm, the tensor is compressed as lower order core one using tucker decomposition. The factor of tucker decomposition can be obtained by left singular value of the  original tensor’s mode3 matrix. The mixture matrix can be estimated by the alternating least square based canonical decomposition of the core tensor. Simulation results show that the proposed algorithm has much lower computational complexity with no performance loss and its operation time reduces 46.44%~76.28% compared with the existing algorithm.