Research On The Key Technology For Optimizing Sensing Cost In Edge-Based Mobile Crowdsensing

With the proliferation of mobile devices and the development of Internet of Things(IoT)technology,mobile crowdsensing as a powerful sensing paradigm,has attracted more and more attention from the industry and academia.Mobile crowdsensing is a large-scale,fine-grained mobile sensing network composed of sensors integrated on smart devices carried by a large number of mobile users.Compared with the traditional sensor network,it has the advantages of wide coverage,low deployment cost and high scalability,which is widely used in traffic navigation,environmental monitoring,etc.With the sensing data becoming more and more fine-grained and complex,the traditional centralized mobile crowdsensing is prone to time delays and increased transmission costs.In view of this,mobile crowdsensing based on edge computing has received more and more attention,which deploys a large number of edge servers between the users and the central server to process and aggregate the sensing data uploaded by the mobile users.In this scenario,how to optimize resource scheduling to reduce the sensing cost has become a key issue.This paper focuses on the key issue of resource scheduling to optimize the sensing cost in edge-based mobile crowdsensing.We perform the research and propose the corresponding strategies from the following two aspects: the edge server configuration and the sensing data uploading.The specific contents are as follows:Since each user can collect multiple types of data in mobile crowdsensing.On purpose of performing data aggregation,the same type of data carried by different users should be uploaded to the same edge server.Therefore,users need to move to different edge servers to upload data according to the type of data carried by them.During this process,the costs are incurred by users(e.g.,the travel distance for uploading data)and edge servers(e.g.,activating server and processing data).In this case,there will be a problem: how to design the edge server activation and configuration strategy to receive the sensing data,so as to minimize the total cost of the data uploading and processing.By formulating this problem as a facility location problem,we propose two algorithms based on linear relaxation and simulated annealing to solve the problem approximately,and analyze the performance of the algorithm theoretically.Finally,we verify the performance of the proposed algorithms with real-world datasets.In addition to reducing the data uploading cost,we also need to consider maintaining the system stability,that is,the stability of the buffer queue in the mobile device(the average production rate of the sensing data should not be much higher than the average uploading rate of the sensing data)and the stability of the buffer queue in the edge server(the average receiving rate of the sensing data should not be much higher than the average processing rate of the sensing data).Therefore,we propose a data uploading strategy based on Lyapunov optimization theory,which reduces the data uploading cost and ensure the system stability.Furthermore,we analyze the performance of the algorithm.Through the simulation based on the real datasets,we conclude that the proposed method can reduce the cost and maintain the system stability compared with the baseline methods.