Blind Detection Of MPSK Signal With Complex Multistate Continuous Hopfield Neural Network

Because of not needing training sequences, Blind equalization and blind detection technique have been received extensive attention and research,.M-ary Phase Shift Keying (MPSK) modulation has been widely used in wireless communication systems as an efficient modulation. To blindly detect MPSK signals,the complex-valued multistate Hopfield neural networks (CVMHNN) is used in this paper.CVMHNN is based on the concept of multivalued logic and adopts the complex-valued neurons, so it has the great advantage in dealing with the multivalue signals. In order to prevent network state from being trapped by a spurious pattern, the continuous phase multistate activation function is designed. Firstly, this article analyzes associative memory Hopfield theory, points out its applications in communications and describes the difference between the associative memory Hopfield and Hopfield for blind detection; then an optimization performance function is constructed, CVMHNN is used to solve the optimization performance function,and a energy function is constructed to prove the stability of CVMHNN with the given activation function. Finally, QPSK,8PSK signals and incomplete Constellation QPSK,8PSK signals are all blindly detected by Hopfield neural network with16PSK activation functions.Computer simulation results show that the used method has better performance for the blind detection of MPSK signals than classical algorithms such as subspace algorithm and so on. Hopfield neural networks have a deeper blind detection capability for blind signal.