NoisyKey: tolerating keyloggers via keystrokes hiding
Stefano Ortolani, Bruno Crispo · Institutional Research Information System (Università degli Studi di Trento) · 2012
Keyloggers are a prominent class of malicious software that surreptitiously logs all the user activity. Traditional approaches aim to eradicate this threat by either preventing or detecting their deployment. In this paper, we take a new perspective to this problem: we explore the possibility of tolerating the presence of a keylogger, while making no assumption on the keylogger internals or the system state. The key idea is to confine the user keystrokes in a noisy event channel flooded with artificially generated activity. Our technique allows legitimate applications to transparently recover the original user keystrokes, while any deployed keylogger is exposed to a stream of data statistically indistinguishable from random noise. We evaluate our solution in realistic settings and prove the soundness of our noise model. We also verify that the overhead introduced is acceptable and has no significant impact on the user experience.