DOA Estimation Methods Based On Cuckoo Search Algorithm

DOA(Direction of Arrival) estimation is an important research content in arraysignal processing，which has been widely applied in radar direction detection andcommunications. Since the late1970s, parameter estimation of signal emerged a largenumber of super-resolution algorithm, where multiple signal classification (MUSIC)and maximum likelihood estimation (ML) are outstanding representatives of thesetwo algorithms. The MUSIC estimation method and ML estimation algorithms can getoptimum performance in theory. But to achieve these two algorithms requires amultidimensional nonlinear search, which calls for a very large amount of calculation.It is a bottleneck hindering the application of these two algorithms.The Cuckoo Search (CS) algorithm is a bionic intelligent optimization algorithmbased on the mechanism of biological reproduction in parasitic cuckoo proposed. Dueto the advantages of the excellent global search and local search capabilities, fastconvergence, less control parameters, etc., this paper chooses CS algorithm tooptimize ML-DOA and MUSIC-DOA estimation.Firstly, the paper introduces some theoretical foundation and mathematicalmodels of DOA parameter estimation, as well as the basic concepts and mathematicalmodels of Cuckoo search. Since the traditional DOA estimation algorithms are basedon the number of the known sources, this paper proposes Information CriterionCriteria and Gaelic Circle Criteria to estimate the number of sources. Both of themcan achieve an effective estimation of the number of signal sources in certaincircumstances of the white noise and snapshots number.Secondly, the paper introduces cuckoo search algorithm to ML-DOA estimation.According to the characteristics of the ML method’s spatial spectrum, the paperimproves the CS search strategy and proposes a fast convergence CS-ML estimationmethod. Through the experimental comparison, the paper proves that maximumlikelihood DOA estimation method based on CS algorithm has a faster solving speed,estimated results are better. Finally, the paper introduces the classic two-dimensional (2-D) MUSICestimation method. For the2-D MUSIC DOA algorithm has a shortcoming of largecalculated amount in the multi-spectral peak searching, the paper proposes a jointDOA estimation method based on2-D MUSIC spectral estimation and isolation nichecuckoo search algorithm. The paper uses isolation niche technique to improve CSalgorithm. And reduce search calculation amount and obtain effective estimate of thearrival angle and pitch angle in consideration of the features of2-D MUSIC spectrumpeak non-linear search. Experimental results show that, MUSIC spectrum estimationalgorithm based on isolation niche cuckoo search estimate performs well.