Signal sources localization using GA preserving genetic diversity based on fitness values

The signal sources localization is a very important study, and many researchers work on this problem by various methods. A genetic algorithm (GA) is probabilistic search method based on a process of the evolution. GA is applied to many real problems, because it has high versatility and search ability. However, in the conventional GA, the reproduction operator causes a lack of genetic diversity, and search efficiency decreases. To solve this problem, the GA in consideration of genetic diversity was proposed. This method can search the optimal solution faster than a conventional GA. However, the parameter which define a degree of variability have to be decided depending on a given problem. In this paper, we propose the new method in which the degree of variability can be adaptively defined based on the fitness values and the spatial distribution of the individuals. The effectiveness of the proposed method is verified by applying it to the signal sources localization using a simple head model.