Spatio-Temporal Privacy-preserving Methods. With the growing advance of location-based services, several spatio-temporal privacy mechanisms have been developed recently (see recent surveys in [7,34,61]). Although the context in mobile crowd sensing is different from location-based services, these mechanisms can be used to address location privacy problems in such scenarios as well. Since location and time are two crucial pieces of information in an effective task management model, applying the existing spatio-temporal privacy-preserving techniques can be challenging. Here, we study some of the applicable methods in MCS task management. Spatial Cloaking In some crowd sensing applications, a perturbed or cloaked location can be used for spatial task management instead of exact locations. Spatial cloaking or perturba- tion hides the participant location inside a cloaked region using spatial transforma- tions [52], generalization (e.g. k-anonymity) [46,101], or a set of dummy locations [54] in order to achieve location privacy. Some MCS applications do not require exact locations (e.g. pollution or weather monitoring), but for the majority of the ap- plications with utility depending on location accuracy, adopting cloaking methods remains a challenge. In our proposed work in Chapter 3, participants share their cloaked location to obtain a set of closest tasks. We developed probabilistic methods to deal with uncertainty for a globally optimized task assignment. Xxxxxx et. al. [48, 49] showed that spatial k-anoymity methods used in location- based services are not directly applicable to Participatory Sensing. Therefore, they proposed that a group of the representative participants ask for spatial tasks from an untrusted server, and share their results with the rest of participants. They would also adjust the spatial regions in queries to make queries independent from the location of other participants. Vu et. al. [97] proposed a spatial cloaking mechanism for Participatory Sensing based on k-anonymity and locality-sensitive hashing (LSH) to preserve both locality and k-anonymity. While most traditional location cloaking methods rely on syntactic privacy models and are subjective to inference attacks, recent works applied more rigorous privacy notion based on differential privacy. The work in [6] proposed a location perturbation method based on a rigorous notion of indistinguishability, which is similar to the differential privacy concept. Another recent work [105] protects the exact locations with differe...
