An Improved MUSIC Algorithm Implemented with High-speed Parallel Optimization for FPGA

This paper proposes an improved MUSIC algorithm with high-speed parallel optimization for FPGA. Although MUSIC algorithm is a high-performance, classic DOA method, it needs estimation and eigenstructure decomposition of covariance matrix, which is time-consuming with high computation cost and not suitable for FPGA implementations. In this paper, the authors present an optimization algorithm without the eigenstructure decomposition of the covariance matrix. This algorithm offers far lower computation cost compared to MUSIC at the expense of little performance decrease. With parallel preprocessing focusing on correlation matrices estimation and spectral peak search, an FPGA implementation is introduced, and proved to be efficient through theoretical analysis, simulation and hardware implementation.