Improvement of the performance of distributed OS-CFAR system by (μ+λ)-ES optimisation

Genetic algorithms (GAs) are algorithms of exploration based on natural selection and on genetic. They are very flexible tools used to optimise very irregular functions, badly conditioned or complexes to calculate. The use of reproduction operators: crossover and mutation, and also the cumulative information prune the search space and generate a set of plausible solutions. Also, other techniques based on the evolutionary strategies (ESs) are proposed in literature as heuristic optimisation techniques. In this work we propose an optimisation of distributed OS-CFAR systems parameters by both a GA and an ES in order to optimise the threshold and also to give a comparison between the two manners to achieve the best performance in detection. The results showed that some improvement had brought by the use of the ES according to the number of sensors in the system, the number of cells in the sensor, the Probability of false alarm (Pfa), and the fusion rule.