Tis-Bad: A Time Series-Based Deobfuscation Algorithm
Location-Based Services (LBSs) have brought along many benefitsto users and service providers in terms of improved quality of existing services and a better user experience. At the same time, location privacy has become one of the most critical concerns from the users’ point of view. One of the existing techniques to protect the users’ location is through Location Obfuscation, which consists of altering the location of the user while still allowing the provider to provide the requested service. Due to the simplicity of some techniques of this kind, they may not offer enough protection against deobfuscation attacks (try to infer the original information from the obfuscated one), but there have not been much work on performance evaluation of them. This work presents a formal definition of a deobfuscation technique for noise-based obfuscation algorithms called TIS-BAD (Time Series - Based Deobfuscation) which implements an exponentially weighted moving average over the obfuscated data to filter the induced noise. In the literature there have been very few efforts to present such formal deobfuscation techniques, being this is one of the main contributions of this work. We evaluate the TIS-BAD algorithm against the Rand and N-Rand obfuscation algorithms, including both location and time scrambling, for straight and non-straight routes. The results show that the TIS-BAD algorithm can filter from 47% to 60% of the induced noise by the obfuscation algorithms, reducing considerably the protection on the users’ location information.