Application of evolution programming for blind equalization

This paper presents an application of evolution programming (EP) for blind equalization (BE). A signal transmitted over a distorting channel needs equalization to be reproduced. BE is a good option when no training signal is available. The finite size of the source alphabet is exploited to implement a block-processing scheme based on EP for designing blind equalizers. The problem reduces to optimizing a multimodal multivariable cost function. The search is made more effective by varying a radix used to represent the species. To ensure a unique solution some constraints are used in the generation process of the offspring.