Relaxed look-ahead pipelined nonlinear channel equalizer

This paper presents a pipelined equalizer for nonlinear channel environment using the relaxed look ahead technique. Nonlinear channel is a known problem that can be treated by artificial neural network (ANN). But, the recursive nature of the adaptation algorithms found in ANN limits their operation speed and high throughput applications require pipelining. The standard look ahead technique is used to pipeline recursive type algorithms and the result often presents a large amount of hardware due to pipelining. In order to minimize this hardware, the relaxed look-ahead technique uses approximations of the look-ahead technique. The combination of these approximations in conjunction with parameter variation can result in a large variety of architectures. This paper will present the relaxed look-ahead technique in a general form and its application to the backpropagation algorithm found in ANN. Simulation results with linear and nonlinear channels will be shown for different pipeline depth.