Study of Imposter Attacks on Novel Fingerprint Dynamics Based Verification System

The rapid momentum in the realization of security solutions, availability of affordable hardware components, and computing devices has led to a tremendous rise in biometric research. However, the threat of spoofing has raised palpable security concerns. In this paper, we examined a recently introduced novel behavioral biometrics technique, namely, fingerprint dynamics. The technique exploits individual’s behavioral characteristics observed from multi-instance finger scan events. The objective of this investigation was to study the spoof resistance capabilities of the fingerprint dynamics-based standalone identity verification system. We used a custom-built hardware unit to collect biometric samples from a total of 50 participants, in an environment that closely mimics the operational scenario. Data collection was done in several sessions per user, spread over a period of seven weeks. We performed an exhaustive analysis of several time-derived features, and selected a combination of best-performing features using genetic algorithm. We also conducted a systematic evaluation using support vector machine and $k$ -nearest neighbor classifiers. We performed a series of verification experiments under three different and practically relevant attack scenarios, namely: 1) combined zero-effort and active imposter; 2) only zero-effort imposter; and 3) only active imposter. We find that the proposed technique exhibits promising results under all the three attack scenarios.