Development of a multidimensional anonymization technique for preserving privacy in participatory sensing system

Abstract

Participatory sensing technology is designed to facilitate community people collect, analyze, and share information for their mutual bene t in a cost-e ective way using smart-phones, camera or other ad-hoc sensing devices. The apparently insensitive information transmitted in plaintext through a lightweight infrastructure of participatory sensing system can be used by an eavesdropper to infer some sensitive information and threaten the privacy of the observer. Su cient number of participants is imperative for the success of participatory sensing. Participation depends a great deal on the assurance of privacy protection. Existing techniques add some uncertainty to the actual observation to achieve anonymity of the participants which, however, diminishes data integrity to an unacceptable extent. A subsetcoding based anonymization technique was proposed in [1] to safeguard observers' location privacy from adversaries while preserving almost loss-less data integrity at the destination server. However, the high computational complexity of that technique O(N!) allowed its use at limited level. In this thesis, we develop an O(N) technique to overcome this limitation. The new technique accommodate variable degree of desired anonymization for the users which eventually enables designing exible incentive schemes for the users. Finally, to the best of our knowledge, we present the rst multi-dimensional privacy preserving scheme that can protect users' privacy at di erent dimensions simultaneously. For example, both location and product association of an observer can be protected. Comprehensive simulation and Android prototype based experiments are carried out to establish the applicability of the proposed schemes.