Channel Equalization Using Dynamic Fuzzy Neural Networks

Channel equalization is a major method for reducing distortion and interference effects on a communication channel. In this paper, channel equalization using soft computing methods is attempted. To be more specific, Dynamic Fuzzy Neural Networks (DFNN) which combines fuzzy rules and neural networks is adopted. The DFNN is functionally equivalent to a Takagi-Sugeno-Kang (TSK) fuzzy system possessing the learning ability of a Radial Basis Function (RBF) neural network. The hidden neurons (rules) of the DFNN can be added and pruned dynamically during the training process based on the significance of each neuron to achieve a compact topology structure. Simulation studies demonstrate that the performance of the DFNN-based equalizer is superior to some other existing equalizers in terms of Bit Error Rate (BER).