Preserving Accuracy and Privacy in Participatory Sensing Systems
Participatory Sensing (PS) is an approach that offers individuals and interest groups the opportunity to contribute to an application using their handheld sensor devices such as smartphones and tablets. These sensor devices are able to sense, collect available data and use cellular and Internet communication infrastructure such as LTE and WiFi to transmit the data to the application server. The application server processes the collected data and makes the data available to the end-users. Participant contributions consist of sensor data, location, date and time. In addition, PS applications usually need to collect additional data about participants such as identity, age, gender and contact. Notwithstanding the numerous benefits the PS approach brought to the application domain, there are two main challenges that threaten the success of PS: data trustworthiness and participant privacy. The additional data collected from participants’ devices are essential to verify the credibility of participants and the accuracy of their contributions. Moreover, these additional participant data are considered private. Thus, ensuring data trustworthiness and accuracy sacrifices the participant privacy, and vice versa. In this thesis, we propose a framework for PS that involves three major schemes to overcome the challenges of accuracy-privacy trade-off. The framework ensures participant contribution data trustworthiness in PS applications, verifies the accuracy of participant contributions in critical situations, and protects participant privacy in critical situations. PS applications are usually open to the public, and receive sensor data from multiple participants. This openness feature of PS applications allows inaccurate and corrupted contributions to affect the quality of the application services negatively. A way of ensuring contribution validity is by evaluating participant reputation values through a designed reputation system. Therefore, we propose a Reputation System to Evaluate Participants (RSEP) to ensure participant contribution data trustworthiness and provide accurate participant contributions. When a crisis occurs, immediate response by rescue personnel is crucial. Decisions for a rescue plan are based solely on data about the crisis from the location. Receiving data from the public could potentially result in corrupted and inaccurate data that will negatively impact the rescue plans. Therefore, we propose a Participant Contribution Trust scheme (PCT) that allows the PS application to verify the accuracy of contributions before sending the data to the crisis response system that requires all available data in order to reach its optimal performance. In critical situations when a crisis occurs, the accuracy-privacy trade-off becomes more complex. Adding more weight to one side needing accurate data, over the other, risking breach of privacy, may become essential due to the specific situation. When a participant is at risk, data accuracy becomes more important than participant privacy. Thus, we propose a Context-Aware Privacy scheme (CAP) that balances the privacy-accuracy trade-off. The CAP scheme eventually provides privacy-preserved data to authorized recipients based on the status of the participants. Depending on the recipient category, their role and policies enforced, a different level of participants’ private data may be received.