LLM-Assisted Optimisation of Multi-RIS Placement and Beamforming in Smart Warehouses

In this paper, we propose an optimization framework for deployment of multiple reconfigurable intelligent surfaces (RISs) to meet the wireless coverage demands for smart warehouses. Specifically, we are the first to formulate a unified network optimization task that jointly considers RIS placement and beamforming to maximize overall network coverage with deterministic channel model to accurately describe the multipath effect for the warehouse. To address this problem, we design a hybrid optimization framework composed of three synergistic modules. (1) A Large Language Model (LLM) serves as a semantic-based planner, generating physically meaningful multi-RIS configurations that jointly specify the placement and beamforming of all RISs, guided by structured prompts and environment-aware embeddings. (2) An evolutionary mutation module performs localized refinements over the given configurations to enhance numerical precision. (3) A Diversity Reflection and Correction (DiRect) module evaluates structural similarity among configurations, ensuring sufficient exploration of the solution space and preventing premature convergence. Extensive simulations validate the effectiveness and robustness of our proposed framework. Compared with traditional heuristics, reinforcement learning methods, and LLM-guided baselines, our approach achieves 10%-15% higher coverage within only 10-20 iterations. The performance consistently scales with the increasing of RIS quantities, element sizes, and remains stable under varying transmitter positions, demonstrating strong adaptability to complex smart warehouse layouts. Overall, the proposed hybrid optimization framework offers a scalable and physically grounded solution for RIS-assisted network deployment optimization in realistic warehouse environments.