Ubiquitous sensing for security in smart homes

Thesis

In recent years, smart home systems have become increasingly popular. With this increased popularity has come a newfound prevalence of IoT devices with varied sensor functionality in peoples' homes. These types of IoT devices are widely considered to be insecure and many problem-specific solutions have been proposed. However, we argue that the widespread presence of diverse sensors in and of itself can—and should—be used to strengthen the security of smart home environments.

In this thesis, we show that the ubiquity of sensors in modern smart homes can improve the overall security of smart home systems, and that it can enable new security functions. To this end, we introduce two security systems based on ubiquitous sensors already available in modern smart homes. The first is a drone detection system that relies only on radio signal strength information that can be provided by most wireless consumer devices. The second is an event verification system for event-driven smart homes that uses physical event signatures from a multitude of sensors to verify the occurrence of physical events in smart home environments. We then analyze how such systems are affected by two advanced types of attackers: an opportunistic attacker who can wait for and predict attack opportunities, and an attacker who can compromise sensor data to create these opportunities themselves. Finally, we demonstrate that contextual information widely available in smart homes can be employed for context-aware sensor selection due to improved situational awareness.

Extrapolating from the insights gained within this thesis, we argue that ubiquitous sensing has significant potential to provide stronger security properties in smart home systems, as it allows for the combination of fine-grained sensor data with high-level event information. Crucially, this added security can be achieved with existing smart home devices—without impacting their original functions.

Authors: Birnbach, S

• DOI: 10.5287/ora-1zpqoe2kr
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English