Scaling up MIMO radar for target detection

TL;DR

This paper investigates target detection in colocated MIMO radar systems by analyzing the asymptotic behavior of a Wald-type detector as the number of antennas increases, providing practical insights for large-scale systems.

Contribution

It introduces an asymptotic analysis of a Wald-type detector in MIMO radar, deriving closed-form false alarm and detection probabilities without prior statistical knowledge.

Findings

Asymptotic detector follows a chi-squared distribution as M→∞

Closed-form expressions for detection probabilities are obtained

Performance matches finite system results for M≥50

Abstract

This work focuses on target detection in a colocated MIMO radar system. Instead of exploiting the classical temporal domain, we propose to explore the spatial dimension (i.e., number of antennas $M$) to derive asymptotic results for the detector. Specifically, we assume no a priori knowledge of the statistics of the autoregressive data generating process and propose to use a mispecified Wald-type detector, which is shown to have an asymptotic $\chi$-squared distribution as $M\to\infty$. Closed-form expressions for the probabilities of false alarm and detection are derived. Numerical results are used to validate the asymptotic analysis in the finite system regime. It turns out that, for the considered scenario, the asymptotic performance is closely matched already for $M\ge 50$.