A privacy preserving optimized intelligent security framework for smart homes using zero trust architecture and explainability

Modern smart home environments include traditional sensors, autonomous robots, connected vehicles, intelligent consumer electronics devices, and wearables, along with context-specific computing and services. The heterogeneity of smart devices, together with their communication protocols, applications, and cloud-hosted services, introduces new vulnerabilities and raises security and privacy concerns. End-to-end foolproof security and data privacy guarantees are currently difficult to achieve and remain an open challenge. This paper proposes POIZE (Privacy, Optimization, Intelligence, ZTA, and Explainability), a comprehensive security framework for modern smart homes. POIZE focuses on preserving privacy, optimizing data flow and processing, using Tiny Machine Learning (TinyML) models for anomaly detection, applying Zero Trust Architecture (ZTA) principles, and providing explainability to end users. The proposed framework combines ZTA principles with novel data processing and computation optimization, while emphasizing explainability, to meet the evolving needs of users and address key security and privacy challenges in smart home environments. Preliminary experimental results demonstrate the feasibility of the proposed framework.