Adaptive and Robust Data Poisoning Detection and Sanitization in Wearable IoT Systems using Large Language Models

2025-11-04

AI Summary

Human activity recognition (HAR) in wearable Internet-of-Things (IoT) systems is highly vulnerable to data poisoning attacks, while conventional defenses rely heavily on large-scale labeled datasets and exhibit poor adaptability. Method: This paper proposes the first large language model (LLM)-driven zero-shot/few-shot detection and purification framework tailored for wearable-device data security. It innovatively integrates role-playing prompting with chain-of-thought reasoning to enable sensor-level anomaly detection and adaptive data restoration—without requiring task-specific training data. Contribution/Results: Extensive experiments demonstrate that the framework significantly outperforms baselines across key metrics—including detection accuracy, purification quality, end-to-end latency, and communication overhead—thereby enhancing the security and robustness of HAR systems in dynamic IoT environments.

Technology Category

• Application Domains: Internet of Things, Sensor Networks & Smart Cities
• Machine Learning: Hardware-aware ML
• Humans and AI: Human-in-the-loop Machine Learning

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

The widespread integration of wearable sensing devices in Internet of Things (IoT) ecosystems, particularly in healthcare, smart homes, and industrial applications, has required robust human activity recognition (HAR) techniques to improve functionality and user experience. Although machine learning models have advanced HAR, they are increasingly susceptible to data poisoning attacks that compromise the data integrity and reliability of these systems. Conventional approaches to defending against such attacks often require extensive task-specific training with large, labeled datasets, which limits adaptability in dynamic IoT environments. This work proposes a novel framework that uses large language models (LLMs) to perform poisoning detection and sanitization in HAR systems, utilizing zero-shot, one-shot, and few-shot learning paradigms. Our approach incorporates extit{role play} prompting, whereby the LLM assumes the role of expert to contextualize and evaluate sensor anomalies, and extit{think step-by-step} reasoning, guiding the LLM to infer poisoning indicators in the raw sensor data and plausible clean alternatives. These strategies minimize reliance on curation of extensive datasets and enable robust, adaptable defense mechanisms in real-time. We perform an extensive evaluation of the framework, quantifying detection accuracy, sanitization quality, latency, and communication cost, thus demonstrating the practicality and effectiveness of LLMs in improving the security and reliability of wearable IoT systems.

Problem

• Detecting data poisoning attacks in wearable IoT human activity recognition systems
• Reducing reliance on large labeled datasets for IoT security defenses
• Improving real-time security and reliability of dynamic wearable IoT environments

Innovation

• LLMs detect data poisoning via role play
• Step-by-step reasoning infers sensor anomalies
• Zero-shot learning enables real-time sanitization