CRB-Rate Tradeoff in RSMA-enabled Near-Field Integrated Multi-Target Sensing and Multi-User Communications

TL;DR

This paper addresses near-field integrated sensing and communications (NF-ISAC) with extremely large antenna arrays by leveraging rate-splitting multiple access (RSMA) and hybrid analog-digital beamforming (HAD) to manage interference across multiple targets and users. It derives the Cramér-Rao bound (CRB) for joint distance and angle sensing, and defines a CRB–rate Pareto boundary to quantify the trade-off between sensing precision and communication throughput, formulating a sensing-centric optimization with rate constraints. A penalty-dual decomposition (PDD) framework with a WMMSE-based reformulation enables joint optimization of fully-connected HAD beamformers and common rate allocations, complemented by a low-complexity two-stage design; the approach is extended to partially-connected architectures. Numerical results demonstrate that RSMA-HAD NF-ISAC approaches the performance of full-digital beamforming with far fewer RF chains, significantly outperforms SDMA and far-field ISAC benchmarks, and yields a superior CRB-rate region, highlighting practical gains for NF multi-target sensing and multi-user communications.

Abstract

Extremely large-scale antenna arrays enhance spectral efficiency and spatial resolution in integrated sensing and communication (ISAC) networks while expanding the Rayleigh distance, triggering a shift from conventional far-field plane waves to near-field (NF) spherical waves. However, full-digital beamforming is infeasible due to the need for dedicated radio frequency (RF) chains. To address this, NF-ISAC with a rate-splitting multiple access (RSMA) scheme is developed for advanced interference management, considering fully-connected and partially-connected hybrid analog and digital (HAD) beamforming architectures. Specifically, the Cramér-Rao bound (CRB) for joint distance and angle sensing is derived, and the achievable performance region between the max-min communication rate and the multi-target CRB is defined. To fully characterize the Pareto boundary of the CRB-rate region, a sensing-centric minimization problem is formulated under communication rate constraints for two HAD beamforming architectures. A penalty dual decomposition (PDD)-based double-loop algorithm is developed to optimize fully-connected HAD beamformers. To reduce computational complexity, a two-stage design algorithm for fully connected HAD beamforming is also proposed. Additionally, the PDD-based double-loop algorithm is extended to the partially-connected HAD architecture. Simulations demonstrate the proposed schemes and algorithms: 1) achieve performance comparable to a fully digital beamformer with fewer RF chains, 2) outperform space division multiple access and far-field ISAC, and 3) yield enhanced CRB-rate trade-off performance.

Figures (5)

• Figure 1: The considered RSMA-aided NF-ISAC networks.
• Figure 2: RCRB versus the minimum rate requirement
• Figure 3: RCRB versus transmit power threshold
• Figure 4: RCRB versus the number of RF chains
• Figure 5: RCRB versus the number of users