Direction of Arrival Estimation Using Genetic Algorithm over MUSIC-Like Spectrum

Due to easy implementation, applicable to a variety of array geometries and ability to resolve closely spaced signals, MUSIC (MUltiple Signal Classification) algorithm has been the most widely studies and has many variants for the directions of arrival (DOA) estimation methods. However, this generality is accompanied with computationally expensive because it involves complex-value computation, especially, in the task of a matrix decomposition. To overcome this problem, the real-value version of MUSIC called MUSIC-like spectrum is reformulated. Peaks in the MUSIC and MUSIC- like spectra represents the DOA estimates. A spectral search for peaks is done by means of the genetic algorithm which offers several benefits over the tradition optimization tools. The performance of the MUSIC-like method and the DOA estimation is evaluated by the comparison of the spatial spectra, simulation times and the accuracy indicators consisting of the root mean squared error and deviation of the DOA estimate.