Evolving complex adaptive IIR structures

A new Evolutionary Algorithm (EA) has been formulated for evolving adaptive complex infinite impulse response (IIR) structures. It is a variation of the standard Genetic Algorithm (GA) in which a new operator called "immigrant" is introduced in addition to the standard genetic operators (crossover and mutation). This operator aims to avoid premature convergence and to increase the search points while maintaining a small population size. Filter stability is ensured by removing any unstable systems from the population as it evolves. The algorithm is applied to the design of a channel equalisation filter and it shows a significantly better mean square error performance (MSE) when compared with an equivalent filter optimised through traditional GAs.