Research Of Blind Equalization Algorithm Based On Neural Network

Blind equalization technology is a new branch of modern adaptive equalization techniques, it compared to the traditional adaptive equalization techniques, it receives signals using only some of the data features from itself to adjust the parameters, thereby eliminating the equalizer by channel caused the non-ideal characteristics between strings, achieve a balanced code of channel purpose. This equilibrium method could not only enhance communication efficiency, but also can reduce communication from a small degree system's complexity.In this paper, first blind equalization techniques and neural network of principle and development status were introduced in detail, and then based on neural network for all sorts of advantages and disadvantages of blind equalization techniques is analyzed. Including high-order neural network has very strong nonlinear transform ability. Also, since the network contains no hidden, so the algorithm convergence speed is fast, and not easy to local minima. However, it is the nonlinear transform ability and order number proportional to the number, and when order, the network weight matrix in the dimension of geometric power pace, this leads to the network using hardware realization up become very difficult. This paper introduced the problems bilinear feedback neural network (BLRNN). This network of nonlinear equations can be approximation, and at the same time, when the order number increase the amount of calculation, the network also won't form of geometry power, which makes its increased to more easily to use hardware to implement. According to the characteristics of blind equalization algorithm (BLRNN) adaptive algorithm, and the traditional constant-modulus algorithm, and success with this algorithm applied to complex domain.Finally to such algorithm by computer in different under the channel is simulated, and the results show that this algorithm is compared with the traditional CMA in performance in a certain degree of improvement.