Research On Adaptive Equalization Algorithm Based On Feed-Forward Neural Network Optimized By Genetic Algorithm

With the development of technology, the modern communication environment has become more complex. Because of the communications often suffer such as channel-fading,inter=symbol interference(ISI),channel interference and side-channel interference which can lead to transmission channel is greatly distorted, causing communications are deteriorating. The Adaptive equalization technology is overcome inter-symbol interference, improve communication quality, reduce the probability of error effective method. To achieve channel equalization, its filters the received signal which demodulate by the Modem.In real communication systems, most of the channel is non-linear model, there is no completely linear channel. The neural network is a typi-cal nonlinear processing system, which can handle complex nonlinear problems,and it has good self-organization, self-learning performance and large-scale parallel processing ability.By using the neural network system to improve equalizer effect, the communication quality will be improved significantly. However the neural network has local search ability, when the unresolved question has more extreme values,will can easy fall into partially smallest, without obtain the global optimum solution. Genetic algorithm is a kind of bionics algorithm,which is based on evolutionary biology and genetic theory. Genetic algorithm is a random searching method with global and parallel characteristics, and it owns powerful robust and global convergent properties.Through summarizes limitations of traditional algorithms, Analyzes the combination of neural network and genetic algorithm, the paper adopts Genetic Algorithm in neural network adaptive equalization by optimizing the structures and weights of these networks, discoved the new algorithms can obtain better neural network adaptive equalization effect,the computer simulations show that the improved algorithm obtains good convergence performance.The main of this paper can be summarized as follows:(1)The drawbacks of the typical adaptive equalization algorithm based on the three-layer feed-forward neural networks are analyzed, according to the deficiency of above algorithm, the paper mainly studies the combination of neural network and genetic algorithm, a new one Genetic Algorithm to optimize neural network weights and structure of the algorithm ideas.(2)According to the advantages of genetic algorithm, Pointed new algorithms of adaptive equalization based on neural network and GA. Both combine the neural network with GA through constructing the fitness function with the cost function of adaptive equalization,and then applies the algorithm in optimizing weights and structures of Feed-forward neural network respectively.By different signal simulation, the results proved new adaptive equilibrium algorithm obtain improved obviously in convergence speed and residual error.(3)Based on standard genetic algorithm to optimize the network structure existing in the long searching time, large memory and genetic operation multifarious, the paper adopts elitism-based compact genetic algorithm to optimize neural network structure. Through computer simulation show that in the linear and non-linear channel, this neural network adaptive equalization algorithm achieves faster convergence speed and smaller residual error.