Research On Participant Selection Method Based On Task Region In Mobile Crowd Sensing

In recent years,with the popularization of personal handheld devices such as smart phones equipped with various types of sensors,and the rapid development of wireless sensor networks and mobile computing technologies,the use of a multi-source-aware device carried by a mobile user to reach the target area of the task to complete the collection of the sensing data has a broad application prospect.Mobile crowd sensing has quietly become a promising new model for promoting the application of urban sensing.In the whole process of mobile crowd sensing,three main processes of task distribution,sensing data collection,and data analysis are required.Two of these processes require the participation of the task server,and the data of the last process is also contributed by the participants,so the participants of the task have an indispensable value status for the mobile crowd sensing.For most mobile crowd sensing applications,the platform has a limited budget for each task,so how to select a group of participants so that they can guarantee the quality of sensing tasks and control the cost of the platform becomes the primary challenge for mobile crowd sensing.For this reason,this dissertation conducts an in-depth study of participant selection methods based on the sensing task region.Considering that sensing tasks in cities are mostly location-based,and their target regions are mostly distributed in hot spots that people often visit,matching the task regions to hot spots in cities is the most important issue.At present,most of the hotspot extraction methods are based on the uniprocessor mode,and the processing time for large-scale trajectory data is long.Meanwhile,it is difficult to meet the sensing tasks that require higher timeliness.Therefore,a hotspot region extraction method based on distributed parallel clustering has been firstly proposed.Firstly,the whole dataset is abstracted in a rectangular region,and the dataset is divided into several partitions with tasks that have almost the same amount by the transformation of the longest dimension of the rectangle.So the local datasets for distributed parallel clustering are constructed.Then the worker servers implement the DBSCAN clustering algorithm for the local partition respectively,and the manager server merges and integrates the local clustering results.The experimental results show that this method can efficiently extract hotspots in cities and the computational rate of clustering analysis is improved to a certain degree.There are a large number of mobile users in the crowd sensing platform,scattered in every corner of the city.However,only online users can receive the platform's sensing tasks.In order to make full use of various user resources in the platform,a participation selection method based on collaborative coverage of the task region has been proposed.It uses the opportunities of online and non-online users to collaborate to accomplish the same sensing task.The method firstly uses the context semantic relationship of the user's movement trajectory to predict the user's future location data.Based on this,the probability of the user's arrival in the task region and the probability of the user's encounter are calculated,and finally the participant is selected according to the user's ability to complete the task.Through experimental analysis,this method can maximize the number of participants under the premise of a certain budget,and ensure the coverage of the target region of the sensing mission.