FPGA Implementation of DOA Estimation Method Based on Polarization Sensitive Array

Lutao Liu,Ying Cao,Muran Guo
DOI: https://doi.org/10.21203/rs.3.rs-1701368/v1
2022-01-01
Abstract:Abstract In this paper, the Field Programmable Gate Arrays (FPGAs) implementation of multiple signal classification (MUSIC) algorithm based on polarization sensitive array is discussed for the first time. Directly embedding the conventional complex-valued MUSIC in FPGA will waste a lot of hardware resources and involve high computational burden. Thus, we propose a real-valued direction of arrival (DOA) estimation method based on polarization sensitive arrays for FPGA implementation in this paper. The proposed method introduces a linear transformation on the received signal, thus simplifying the subsequent calculations of the polarization MUSIC algorithm. In addition, the whole FPGA implementation scheme using the proposed real-valued MUSIC based on distributed polarization-sensitive arrays is also developed in this paper. The proposed scheme releases the computational burden via exploiting the parallel calculation of the covariance matrix, the parallel Jacobi algorithm of multiple cleaning of eigenvalue decomposition (EVD), the multi-level spectral peak search method, and the pipelining of each module. The multi-level and multi-group cleaning in the EVD module reuse the resources occupied by a group of cleaning, and the spectral peaks are serially searched in the peak search module. Consequently, compared with the complex-valued MUSIC, the hardware resource consumption and computation burden is dramatically reduced in the proposed scheme.
What problem does this paper attempt to address?