Analysis of location privacy risk in a plain-text communication based Participatory Sensing System using subset coding and mix network

Success of participatory sensing, where participants share information of their environment to an application server using readily available mobile sensor devices such as smart phones, depends on mitigating the location privacy risk as identity of the participants need to be relayed to facilitate rewards. Recently, we proposed a plain-text communication based location anonymization scheme for participatory sensing without compromising data integrity at the application server using our novel subset coding technique. To improve decoding efficiency, instead of using any random k-anonymization, the scheme has to rely on an anonymization server to which an observer communicates via HP3-based mix network so that the identity of the observer can be protected. Introduction of such a network, however, may increase the risk of location privacy when an adversary within the mix network colludes with other types of existing adversaries. This paper presents a detail analysis of such risk with extensive analytical and simulation results to confirm that the risk depends on a number of system parameters and, more importantly, by increasing the network-friends per user in the HP3-based mix network, the risk can be mitigated without increasing the computational overhead or compromising other goals of participatory sensing such as high data integrity at the application server.