Abstract
This letter presents a novel hybrid automatic repeat request (HARQ) framework tailored for extra-large multiple-input multiple-output (XL-MIMO) systems, which leverages spatial visibility region (VR) structures to enhance retransmission efficiency. Traditional HARQ protocols rely heavily on time diversity, which limits responsiveness under latency constraints and overlooks the spatial richness of XL arrays. To overcome this, we propose a VR-aware HARQ strategy that enables spatially disjoint retransmissions within a single coherence interval. By mapping and classifying user-specific VRs, the framework supports both fixed and adaptive subarray selection under Chase combining (CC) and incremental redundancy (IR) HARQ mechanisms. A lightweight scheduling mechanism is developed to reduce latency, energy consumption, and redundant retransmissions while improving decoding success rates. Simulation analysis validates the effectiveness of the proposed approach, demonstrating significant gains in HARQ success rate, average number of retransmissions, and secure throughput over conventional HARQ schemes.