A new blind deconvolution algorithm for SIMO channel based on neural network

This paper presents a novel approach for blind deconvolution of SIMO channel, and a MISO neural network can easily implement the proposed algorithm. Contrary to gradient-based algorithms, there are no step size parameters to choose. If the number of observed signals is not less than 2, the source signal can usually be separated from the observed mixtures. The new algorithm converges very fast. Furthermore, the extra assumptions proposed in Luengo´s and Zhang´s algorithms about the source signal are not necessary in this paper. The experiments demonstrate the good performance of our algorithm and show that the proposed algorithm is robust in noise environment.