A cyber-physical approach to secret key generation in smart environments

Encrypted communication in wireless sensor networks oftentimes requires additional randomness and frequent re-keying in order to avoid known-plain text attacks. Conventional approaches for shared secret generation suffer however from various disadvantages, such as necessity of a trusted third party, protocol scalability, and especially, the computational resources needed for performance-demanding public-key protocols. To appropriately respond to the increasing disproportions between a computationally powerful adversary and lightweight wireless devices, a cyber-physical approach has recently attracted much attention. The general idea is to leverage the properties of the physical world and include them in a design of lightweight security protocols. Especially valuable physical property is the erratic and unpredictable nature of multi-path signal propagation which has already shown itself as a rich source of randomness. This work presents a new cyber-physical approach in order to make secure wireless sensor communications and proposes a secret key extraction algorithm that leverages signal strength fluctuations resulting from dynamic physical environments, e.g. environments experiencing human movements. In particular, this work presents a systematic experimental evaluation by using a real-world sensor network, and analyzes the impact of different moving patterns on legitimate devices and an eavesdropper. Finally, this work quantifies the main factors that influence the key establishment algorithm and propose a protocol which allows secret sharing in an effective and efficient way.