EIGENSTRUCTURE ANALYSIS OF PULSE INTEGRATION UNDER COVARIANCE MISMATCH IN SURVEILLANCE RADAR

Abstract

Detection of weak targets in surveillance radar critically relies on the effectiveness of pulse integration under statistical uncertainty. In practice, covariance mismatch between assumed and true models can significantly degrade performance, yet its impact remains insufficiently characterized within a unified framework. This article develops an eigenstructure analysis of likelihood-ratio-based detectors under covariance mismatch. By representing the problem in the eigenbasis of the assumed covariance, the resulting test statistic admits an explicit mode-wise decomposition that separates eigenvalue distortion and subspace misalignment. This formulation unifies coherent, noncoherent, and hybrid integration within a common framework. Analysis and simulations under both parameter and structural mismatch show that performance is governed by effective dimensionality and principal-angle structure, leading to practical design insights into robustness–efficiency trade-offs. The proposed framework provides a principled basis for analyzing and designing eigenstructure-based radar detectors under covariance mismatch.