MALLM: Multi-Agent Decision-Making with LLMs for Multi-User Edge-Sensor Environments

Multi-user environments present significant challenges in coordinating diverse preferences and resolving conflicts around shared resources. Single-agent systems struggle to balance individual needs with collective objectives. We introduce MALLM, a novel framework that deploys personalized LLM-based agents for each user on edge devices, which integrates multi-sensor data fusion with a structured multi-agent decision-making mechanism, processing all data locally for enhanced privacy. Our edge-computing architecture enables real-time deliberation through evidence-based argumentation and consensus formation algorithms. The system continuously refines user profiles through sensor data while managing computational resources efficiently. We evaluate MALLM in a real-world 30-day deployment with four users, demonstrating 87.3% user satisfaction compared to 62.5% for majority voting approaches, while achieving 94.1% safety scores and 91.7% adaptability.