Abstract:

In this paper, niche genetic algorithms are used to optimize the performance of a distributed 3 sensor OS CFAR detection system with different reference lengths and signal to noise ratios. A set of quasi optimum results is given and analyzed, which proves that niche genetic algorithms are efficient for this optimization. The influences of the nonidentity in environment on the performance of the distributed OS CFAR detector for different fusion rules are discussed subsequently. The results indicate that when OR fusion is employed, the system is robust to the nonidentical variety of detection environment, while 2 of 3 and AND fusion are sensitive in the same situation.

Key words: genetic algorithms, niche, nonidentical environments, distributed CFAR detection, data fusion