Hybrid Beamforming in I2V Near-Field ISAC Systems for HRLLC
Abstract
In this paper, we investigate infrastructure-to-vehicle (I2V) multiple-input multiple-output (MIMO) integrated sensing and communication (ISAC) systems employing large-scale antenna arrays. Specifically, we consider time-varying (TV) near-field (NF) channel models to effectively address high-mobility and mitigate beam misalignment, which commonly occur for vehicles in NF regions. To achieve high energy efficiency and satisfy hyper-reliable and low-latency communication (HRLLC) requirements, we develop hybrid beamforming (HB) frameworks at a roadside unit (RSU) in time division duplexing under finite blocklength (FBL) transmissions. Our goal is to design hybrid beamformers at the RSU for both the downlink (DL) and uplink (UL) to maximize the spectral efficiencies (SEs) over the DL and UL under minimum requirements for sensing. Due to the FBL transmissions, penalty terms are introduced to the conventional SE expressions, leading to non-convex optimization problems. To address this challenge, we first derive tractable lower bounds on the DL and UL SEs and propose alternating optimization frameworks to design hybrid beamformers at the RSU. Numerical results verify that the proposed HB with a few radio frequency chains achieves performance close to that of a fully digital precoder and demonstrate fundamental trade-offs between SEs and sensing performance of I2V ISAC systems under HRLLC requirements.