Research On Task Distribution Method For Mobile Crowd Sensing Based On Association Relationship

With the rapid development of wireless sensor networks and the popularity of various portable smart devices,a new data sensing mode,Mobile Crowd Sensing(MCS),has risen rapidly.This mode has gradually become the core of mobile computing.Users who participate in the mobile crowd sensing collaborate and interact through the Mobile Internet to form a large-scale sense network of mobile crowd sensing.In intelligent community services such as smart city or urban noise perception,the mobile crowd sensing network can be used to complete large-scale and complex social sensing tasks.Among them,task distribution is an important part of mobile crowd sensing.There are many problems with the task distribution process,such as complex perception environment and a small number of users participating in the sensing.How to enable the sensing user to receive and execute the perceived task quickly and with low overhead is an important issue worth studying now.The main work of this paper includes the following aspects:(1)The social attributes of the sensing users are quantified as a measure of the relationship between users,and an effective score is set for the user.Based on this,an effective user calculation(EUC)based on social attributes and effective user calculation is proposed.The mechanism filters users according to the attributes of the task.From the perspective of the user,EUC considers the sociality of the user,and the relevant information is transmitted by the user's social network to increase the number of effective users of the platform;From the perspective of the platform,EUC can dynamically adjust users' valid points according to the receiving and submission of tasks,so as to guarantee the effective users of the whole system.(2)The historical trajectory data of users in the above EUC mechanism is mined,and an adaptive variable order Markov estimation algorithm(AVOM)is proposed.The algorithm models the application scenario based on the location,and discovers the association relationship between the users by sensing the historical track of the user performing the task.The activity degree of the user performing the task at each location is calculated,and the value of the activity is calculated according to the obtained activity value.The similarity between users,the value of which is used as a measure of the relationship between users,and the users with a high degree of association(that is,high similarity)are divided into the same cluster.Based on the improvement of Markov model,the AVOM estimation algorithm is proposed to estimate the location of a class of users with high correlation.The experimental part verifies the effectiveness of the algorithm based on a real data set.