Research Programme

Context Aware Security for Smart Homes

(2020-2024)

Network Traffic Analysis Anomaly Detection DIY IoT Sensor Fusion Behavioral Patterns

Internet of Things (IoT) Infrastructure / Systems (IS) Security (S) Human Computer Interaction (HCI)

Smart homes present unique security challenges because attacks can manifest in both cyber and physical domains simultaneously. Traditional Network Traffic Analysis alone cannot detect threats that manipulate physical sensor data such as temperature, humidity, light, sound, and vibration. This project addresses IoT security vulnerabilities by combining network traffic analysis with independent sensor observations to identify suspicious behavioural patterns that would otherwise go undetected. The research pursues three main objectives: reviewing existing cyber-physical anomaly detection techniques, learning smart home behavioural patterns using distributed multi-purpose sensors, and detecting anomalies by correlating network traffic with independent sensor observations. The team develops low-cost Do-It-Yourself IoT sensor nodes that learn expected behavioural signatures across devices and occupants, providing an affordable and extensible monitoring solution for residential environments.

By correlating physical observations with network traffic analysis, the system detects attempted compromises even when attackers spoof smart plugs or block telemetry. This layered approach ensures that cyber-physical attacks which evade a single monitoring method are still caught by the independent sensor layer, providing a more robust security posture for connected homes.

The project has produced publicly available datasets for cyber-physical anomaly detection in smart homes. Along with open-source tools and sensor fusion frameworks, these resources enable other researchers to replicate and extend the approach. The work contributes to a broader understanding of how layered, context-aware monitoring can strengthen residential IoT security.