Multi-distance Based Location Privacy Protection In Spatial Crowdsourcing

Spatial crowdsourcing(SC)is a novel platform,which has appeared re-cently,that people can issue spatiotemporal tasks to ask for help and solve oth-ers' task as workers on this platform.All in all,SC has brought new opportu-nities for different social men(i.e.,employment).It is also a great revolution for social life and SC speeds up the tempo of modern living.However,there are some severe privacy problems in the SC system.Especially,location privacy is paid more attention from researchers and industry professionals.In the pa-per,we propose a framework consists of different distance computation-based location privacy protection protocols in the SC system,namely EMB-L3P,MMB-L3P,CDB-L3P and MDB-L3P,which mainly adopts homomorphic encryption(HE)and the prefix membership.In order to eliminate inefficiency caused by weakness of the encryption method,we also construct KD-tree which is a spa-tial index for encrypted locations.Finally,we do security analysis to show that our protocols can prevent external powerful adversaries and privileged insiders from obtaining participants' location privacy.Then,the performance evaluation compares computation and communication overheads between protocols in our paper,so the following conclusions are showed:1)Our proposed protocols are feasibility and superiority over existed schemes which applies differential pri-vacy(DP)to protect location privacy in term of assignment failure rate.2)Apart from various physical application environments,these four protocols have own advantages against different plaintext's length,different number of workers in each BS and different number of worker candidates.Two overheads of EMB-L3P is invariable according to plaintext's length,while MMB-L3P and CDB-L3P are not applied to the long length of the plaintext.With increases of the ratio between number of workers in each BS and number of worker candidates,two overheads MMB-L3P and CDB-L3P are lower than that of EMB-L3P.In this paper,our main innovations can be be summarized as follows:·Different from DP-based location privacy protection protocols of other pa-pers,our protocols mainly adopt HE and the prefix membership to en-crypt worker and task locations,which have lower assginment failure rate in the phase of task assignment.Thus,locations are in the form of the ci-phertext to transmit,store and compute in the whole process of the system operation without decryption.Moreover,in order to improve efficiency of the SC system,we also construct KD-tree to index encrypted locations.CSP finds an adequate worker group with the KD-tree.Finally,according to computed distances between one task location and the corresponding worker locations,the SC-server gets the worker candidates.·Our proposed location privacy protection protocols are based on various distance computation(i.e.,Euclidean Metric,Manhattan Distance,Cheby-shev Distance and Minkowski Distance).Different protocols have differ-ent application scenarios.Especially,the MMB-L3P is applied in comput-ing the shortest path in the city where buildings are laid out in square blocks and straight streets intersect at right angles.The CDB-L3P is the most suitable choice when task locations submitted by requesters lie in the area which workers cannot travel to.The application scenario of EMB-L3P is the city where the form streets are diverse and buildings are laid out sparsely.The MDB-L3P is a generalization of the Euclidean metric and the Manhattan distance.