FPGA-Based Adaptive Covariance Completion ESPRIT for Real-Time Direction-of-Arrival Estimation

Abstract

This paper presents an FPGA-oriented adaptive covariance completion ESPRIT algorithm for real-time direction-of-arrival (DOA) estimation. The proposed method reconstructs an enhanced covariance matrix from partial covariance information using an adaptive correction mechanism, improving signal subspace estimation while preserving the rotational invariance property of ESPRIT. By avoiding spectral search and reducing covariance reconstruction errors, the algorithm achieves low computational complexity and is well suited for hardware implementation. An optimized FPGA architecture employing parallel and pipelined processing is developed to accelerate covariance computation, subspace estimation, and rotational operator evaluation for real-time operation. The proposed approach is validated through MATLAB simulations under varying signal-to-noise ratios and snapshot conditions, followed by FPGA implementation to evaluate hardware performance. Experimental results demonstrate accurate DOA estimation with reduced latency, efficient resource utilization, and high throughput, confirming the suitability of the proposed design for real-time radar, wireless communication, and smart antenna applications.