Efficient Decoders for Sensing Subspace Code

TL;DR

This paper introduces efficient quadratic-complexity decoding algorithms for sensing subspace codes used in DoA estimation, balancing error performance and computational load for 6G sensing systems.

Contribution

It proposes novel decoding algorithms that reduce complexity from cubic to quadratic for sensing subspace codes, with tunable parameters for performance tradeoffs.

Findings

Decoders achieve near MAP performance at higher complexity levels.

Algorithms significantly reduce decoding complexity from cubic to quadratic.

Performance approaches MAP as complexity increases.

Abstract

Sparse antenna array sensing of source/target via direction of arrival (DoA) estimation motivates design of the sensing framework in joint communication and sensing (JCAS) systems for sixth generation (6G) communication systems. Recently, it is established by Mahdavifar, Rajam\"{a}ki, and Pal that array geometry of sparse arrays has fundamental connections with the design of subspace codes in coding theory. This was then utilized to design efficient \textit{sensing subspace codes} that estimate the DoA with good resolution. Specifically, the Bose-Chowla sensing subspace code provides near optimal code design for unique DoA estimation with tight theoretical upper bound on the error performance. However, the currently known decoder for these codes, to estimate the DoA, is a traditional \textit{Maximum-a-Posterior (MAP) decoder} with complexity that is cubic with the number of antennas. In this work, we propose novel efficient decoding algorithms for sensing subspace codes, that reduce the complexity down to quadratic while providing new knobs to tune in order to tradeoff complexity with error performance. The decoders are further evaluated for their performance via Monte Carlo simulations for a range of SNRs demonstrating promising performance that smoothly approaches the MAP performance as the complexity grows from quadratic to cubic in the number of antennas.