A FPGA-based implementation of MUSIC for centrosymmetric circular array

This paper presents a FPGA-based implementation of multiple signal classification (MUSIC) algorithm for centrosymmetric circular array. To reduce the computational complexity of MUSIC, an improved technique, which exploits the inherent properties of the centrosymmetric array, is proposed. Using this technique, a completely real-valued MUSIC is developed; thus less area and computing time are required. An efficient systolic array is proposed to solve the FPGA-based large matrix eigenvalue decomposition (EVD) problem. This architecture needs only about 25% of the area required to implement BLV array. Also, for the implementation of real-time spatial spectrum computation, a 1-D systolic array with high throughput is proposed.