A Decentralized Zero-Trust Framework for Privacy-Preserving Health Data Collection in AIoT-Enabled Wearables

Abstract

The rapid expansion of the Artificial Intelligence of Things (AIoT) has transformed remote patient monitoring by enabling continuous, real-time physiological data collection. However, traditional centralized architectures face critical challenges regarding data privacy, high latency, and the vulnerability of "implicit trust" models. This paper proposes a Decentralized Zero-Trust Framework (DZTF) specifically designed for wearable ecosystems to ensure secure and privacy-preserving data management.
Our framework introduces a multi-layered security approach that shifts data processing from the cloud to the Edge, utilizing Federated Learning (FL) to perform local inference without exposing raw patient data. To eliminate lateral threat movement, we implement a Continuous Authentication protocol based on behavioral biometrics and device health metrics, enforcing a "Never Trust, Always Verify" policy. Furthermore, a Sharding-based Blockchain ledger is utilized to maintain immutable access logs and trust scores, ensuring transparency without compromising storage efficiency.
The proposed DZTF was evaluated using a high-fidelity simulation of 10,000 IoT nodes. Experimental results demonstrate that the framework achieves an 85.9% reduction in latency and a 62% decrease in network bandwidth consumption compared to standard cloud-centric models. Additionally, the integration of differential privacy and zero-trust controls enhanced security resilience, reducing data exposure risks by over 13x. These findings confirm that the DZTF provides a scalable, efficient, and robust solution for the next generation of secure healthcare AIoT systems.