| Résumé | Smart homes face evolving cybersecurity threats, requiring adaptive defense strategies. Existing frameworks for defending smart homes provide static optimization approaches for security controls that fail to address the dynamic nature of threats and often overlook usability factors impacting user adoption. This work presents a framework that outperforms static security optimization approaches, reducing successful attacks by 7 0 - 7 8 % across a diverse user profile portfolio while maintaining acceptable usability levels. The framework uses time-dependent modeling that adapts to evolving threats and varying user preferences. We incorporate realistic dynamics of smart home security by modeling attack patterns with varying temporal characteristics and security control effectiveness. The framework quantifies usability impacts through user satisfaction, technical capability, and compliance rates. We experiment across six user profiles (security-focused, usability-focused, balanced, elderly, tech-enthusiast, and family home) to demonstrate that the framework consistently finds optimal security-usability tradeoff positions compared to static optimization. For most profiles, the framework improvement in overall reward metrics despite moderate decreases in user satisfaction, showcasing its ability to balance security with usability constraints. The framework is a significant advancement in smart home security that provides adequate protection and maintains high security standards while preserving essential usability. |
|---|
