A Reliable Multi-task Allocation Based on Reverse Auction for Mobile Crowdsensing

Mobile crowdsensing (MCS) has been widely studied with the popularization of mobile intelligent devices, while a large number of sensing data can be obtained by using existing mobile devices. Multi-task allocation is an important problem for MCS. In this paper, we propose a highly reliable multi-task allocation (RMA) framework that considers the offline stage and online stage. In the offline stage, we propose a network flow model based on reverse auction, while adding the bid price compensation (BPC) strategy to ensure workers we recruit are reliable. In the online stage, we propose the asynchronous k-secretary strategy to achieve dynamic real-time recruitment of workers and prove the competitive ratio is 1/e(k). Experimental results over the Rome taxi track data set show that our two-stage framework has better coverage and budget consumption, as well as good computational efficiency.