The MUSIC Spectral Peak Searching Based On Genetic Algorithms And Its FPGA Implementation

In modern electronic warfares, the requirment for accurate and efficient DOAestimation is becoming increasingly high. Among the state-of-the-art methods, theMUSIC algorithm which shows high accuracy and good resolution performance hasbeen applied to a variety of applications to sovle the problem of DOA estimates.Compared with traditional search-based techniques, the genetic algorithm (GA) hasthe advantages of low computational complexity, mplicit parallelism characteristicsand excellent real-time performance, which makes it suitable for the peak search ofMUSIC spectrum.First of all, the coarse-grained parallel genetic algorithm (CPGA) is chosen asthe basic framework of the peak search. After that, the improved coarse-grainedparallel genetic algorithm (ICPGA) which sets different crossover and mutationprobability for the different subpopulations is proposed, which improves thepopulation diversity and reduces the phenomenon of premature convergence.Then, orthogonal design, analysis of variance and F-test method are adopted toanalyze peak search results, by which the input signal parameters which are of greatimportance to the result and the GA parameters are picked out. Based on this,simulation experiments are made to get the optimal parameters of the ICPGA andcomplete the peak search of the singal-source MUSIC spectrum. With multi-sourceMUSIC spectral peak, the orthogonal projection pretransformation is utilized toeliminate the peak using ICPGA. Then peaks of MUSIC is successively searchedand removed, giving all of the spectral peaks that regarding all the source DOAs.Finally, the project design of the FPGA implementation for MUSIC spectralpeak search using ICPGA is described, which accomplishes the design andsimulation of the FPGA implementation for MUSIC algorithm and CPGA.According to the overall simulation results, the system works well and meets thedesigns requirements.