A multi-species genetic algorithm applied to radar scattering centre identification in three-dimensions

Realistic radar target models are required for use in a missile target simulation system. A multi species genetic algorithm is used to extract a radar cross section model from high resolution three dimensional radar target images by identifying the dominant scattering centres. A typical image may require in excess of 2 Gbytes of storage space. Data retrieval overheads on small systems render iterative methods of model extraction impractical. This method exploits the niche finding characteristics of the multiple species in the genetic algorithm to simultaneously identify multiple scattering centres. This significantly reduces the data access requirements with only a moderate increase in processing overhead.