Fingerprinting Smartphones Remotely via Sensors Data

Abstract

Remote fingerprinting of physical devices has already been shown to be feasible, using tiny deviations in the clock frequencies of fingerprinted machines. Both TCP and ICMP timestamps provided the basis for clock skew estimations. In the context of the emerging urban-sensing paradigm, we note the potential availability of logged sensors data, typically comprising timestamps, and probably labelled with external time reference information. This observation leads to the possible exploit of that new timestamps source to identify individual mobile nodes, and compromise participating users’ privacy. Our experiments verify this conclusion, confirming the earlier results published on the same line of research, and pull the attention to a privacy breach that could be mounted remotely, offline and in non-real time, on logs of sensors data with necessary time reference labels.