LAUSR

The emergence of mobile crowd sensing (MCS) platforms makes it possible to collect data in a time and cost efficient manner. However, one of the challenges in MCS systems is obtaining reliable information especially in the presence of impersonators who could provide false reports without getting detected. User recruitment mechanisms adopted in MCS systems hire users such that the Quality of Information (QoI) of the submitted sensing reports is maximized. Despite that, there is still a risk of recruiting impersonators to the system. This problem is even more prominent in the case of continuous mobile sensing tasks, since multiple false sensing reports could be submitted by the impersonator during the sensing period, which can impact the quality of the sensing outcome and hence the performance of the system. Therefore, to ensure the reliability of the continuously submitted data, a more robust recruitment mechanism, which detects and eliminates impersonators during the sensing task, is needed. This work proposes a biometrics-based behavioral trust framework that can support a reliable recruitment process in continuous MCS tasks. Behavioral biometrics are unique behavioral traits that can be used to profile users based on how they naturally perform a specific activity. By leveraging machine learning techniques, these behavioral traits can be used in order to detect impersonators in the system. In this work, a unique model for each MCS worker is built based on their unique interaction patterns with the smartphone’s touching screen. The proposed approach integrates the trained machine learning models with a dynamic continuous recruitment system, which continuously monitors the QoI of the submitted sensing reports and changes the recruited participants as needed. Simulation results of the proposed approach show its efficacy in detecting and eliminating impersonators in continuous sensing recruitment.