Geometric Port Selection in CUMA Systems

Abstract

Compact ultra-massive antenna-array (CUMA) is a novel multiple access technology built on the fluid antenna system (FAS) concept, offering an improved scheme over fluid antenna multiple access (FAMA) that can support massive connectivity on the same physical channel without the need of precoding and interference cancellation. By employing a simple port-selection mechanism that leverages random channel superposition, CUMA can suppress inter-user interference while keeping hardware costs low. Nevertheless, its ad-hoc port-selection strategy leaves considerable room for optimization. In this work, we revisit CUMA and propose two adaptive single-RF port-selection schemes that retain its simplicity while significantly enhancing performance. The first one, referred to as exact optimal half-space (EOHS), dynamically selects the projection direction that maximizes the instantaneous signal build-up across active ports. To reduce complexity while preserving most of the gains, we furthermore introduce a principal component analysis (PCA)-based scheme, which aligns port partitioning with the dominant statistical direction of per-port channel vectors. This method yields a closed-form low-complexity solution, complemented by a tractable analytical framework that provides a closed-form expression for the signal-to-interference ratio (SIR) probability density function (PDF). Simulation results corroborate the analysis, demonstrating that both EOHS and PCA consistently outperform conventional CUMA across diverse user densities, port counts, and FAS aperture sizes. Notably, PCA achieves performance close to EOHS at a fraction of the computational cost. The proposed schemes scale effectively to large-user regimes, offering a compelling complexity-performance trade-off for next-generation multiple access systems.