Multimodel adaptive filtering with genetically determined varying model space

It is well known that for the linear filtering problem with unknown time-invariant or time-varying parameters i.e. the adaptive filtering problem, the multimodel adaptive filter (MMAF) based on the Lainiotis (1973) partitioning theorem is the best solution. It is also known that genetic algorithms (GA) are one of the best methods for searching and optimisation. A new method which combines the effectiveness of MMAF and GAs robustness has been developed. The experiments´ results, obtaining from the performance of this new method to a specific linear model with one unknown time-invariant parameter, indicate that the proposed method provides better estimation of the unknown parameter compared to that of a single MMAF