Conventional token-based authentication methods between wireless devices rely heavily on manual user operation, requiring users to manually input long passwords with a number of random characters for "added security". This is often inconvenient for a user who wishes to log into multiple devices, for instance. Custom passwords are also often short and contain repeating characters, making them easily compromisable due to their predictability. By contrast, context-based pairing eliminates the need for manual password input by exploiting the surrounding context of the authentication system to enable automatic, zero-interaction pairing. A caveat of state-of-the-art context-based-pairing schemes, however, is their susceptibility to wireless exploitation in the form of insider threats, or brute force attacks by co-present adversaries. Similarly, biometric systems enable user identification and authentication without the use of passwords using a user’s physiological or behavioral traits to only allow access to the intended user, although biometric systems are as (in)secure as token-based authentication and are highly error-tolerant due to the inherently noisy nature of biometric templates. Therefore, we propose SIENNA, an inSIder rEsistaNt coNtext-based pAiring scheme that (i) exploits human respiratory signals derived from the mechanics of chest displacement and (ii) merges the fuzzy commitment scheme, the JADE-ICA algorithm, and friendly jamming techniques to prevent insider threats on obstructive sleep apnea (OSA) screening systems. The results show that SIENNA achieves reliable device pairing in environments with high levels of signal noise, such as a home setting consisting of multiple, freely-moving individuals in the background. It also prevents unauthorized receivers from retrieving the secret key, regardless of their location and despite their knowledge of the user’s respiration patterns. Additional work presented in this thesis assesses the feasibility of SIENNA to be deployed at a smaller scale, similarly utilizing the respiratory biometric as the context.

Biography