On-Device LLM for Context-Aware Wi-Fi Roaming

2025-05-07

AI Summary

Traditional Wi-Fi roaming strategies suffer from sticky connections or excessive handoffs in dynamic mobile scenarios, struggling to balance stability and signal quality. To address this, we propose the first lightweight large language model (LLM)-driven roaming framework tailored for resource-constrained end devices. Our method employs a cross-layer LLM architecture that performs environment-aware access point selection and dynamically adapts handoff thresholds at the application layer—marking the first integration of high-level reasoning into real-time physical/link-layer control. Leveraging chain-of-thought prompting, parameter-efficient fine-tuning, model quantization, and structured context fusion, the framework achieves significant improvements over heuristic and deep reinforcement learning baselines across diverse indoor and outdoor datasets. Experimental results demonstrate enhanced roaming stability and superior trade-off optimization between connection reliability and signal quality.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Intelligent Robots: Learning & Optimization for ROB

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Wireless roaming is a critical yet challenging task for maintaining seamless connectivity in dynamic mobile environments. Conventional threshold-based or heuristic schemes often fail, leading to either sticky or excessive handovers. We introduce the first cross-layer use of an on-device large language model (LLM): high-level reasoning in the application layer that issues real-time actions executed in the PHY/MAC stack. The LLM addresses two tasks: (i) context-aware AP selection, where structured prompts fuse environmental cues (e.g., location, time) to choose the best BSSID; and (ii) dynamic threshold adjustment, where the model adaptively decides when to roam. To satisfy the tight latency and resource budgets of edge hardware, we apply a suite of optimizations-chain-of-thought prompting, parameter-efficient fine-tuning, and quantization. Experiments on indoor and outdoor datasets show that our approach surpasses legacy heuristics and DRL baselines, achieving a strong balance between roaming stability and signal quality. These findings underscore the promise of application-layer LLM reasoning for lower-layer wireless control in future edge systems.

Problem

• Improving Wi-Fi roaming stability and signal quality
• Replacing heuristic AP selection with LLM-based reasoning
• Optimizing on-device LLM for low-latency wireless control

Innovation

• On-device LLM enables cross-layer Wi-Fi roaming
• Context-aware AP selection using structured prompts
• Optimized LLM with efficient fine-tuning and quantization