FPGA-Oriented Nystrom-Enhanced IMUSIC Algorithm for Accurate and Low-Complexity DOA Estimation

Abstract

This paper presents a Nystr¨ om-Enhanced Iterative Multiple Signal Classification (NE-IMUSIC) algorithm for accu rate and low-complexity direction-of-arrival (DOA) estimation under practical operating conditions. By exploiting Nystr¨ om based covariance approximation, the proposed method signifi cantly reduces computational complexity while preserving the dominant signal subspace, enabling efficient FPGA implementa tion through parallelism, pipelining, and fixed-point arithmetic. The proposed architecture is validated using extensive Monte Carlo experiments with a uniform linear array, where perfor mance is evaluated in terms of root mean square error (RMSE) under varying signal-to-noise ratios and snapshot conditions. Simulation results demonstrate that the proposed NE-IMUSIC algorithm achieves RMSE performance that approaches the Cram´ er–Rao bound at moderate to high signal-to-noise ratio (SNR) levels. This indicates that the proposed approach maintains high estimation accuracy while significantly reducing computa tional complexity through the use of the Nystr¨ om method for covariance matrix approximation.

Keywords