Task Assignment Model Of Mobile Crowd Sensing Oriented Requirements

In recent years,with the rapid development of computer,integrated circuit and communication technology,the mobile intelligent terminal has integrated a wealth of sensor equipment,such as GPS,accelerometer,gravity sensor,gyroscope,electronic compass,light distance sensor,microphone,camera and so on,which makes it have rich sensing ability.With the rapid development of China's economy,the number of people with mobile intelligent terminals is increasing,which provides the conditions for the acquisition of sensing data in a large range and low-level.Mobile Crowd Sensing(MCS)refers to the use of mobile devices to collect,analyze and share the sensing information and data.With the popularity of mobile devices,mobile crowd sensing technology has been paid more and more attention,and it has been widely used in various applications,such as traffic monitoring,environmental monitoring and mobile social recommendation.In the mobile crowd sensing system,the sensing tasks have many types,wide range and large quantity.Facing different task demands,it is the key to assign tasks to the right users to improve the matching degree between tasks and users,and improve the efficiency of task allocation.The tasks released by the perception platform in crowd sensing are multi-class,and the effectiveness of users to complete different tasks is different.Although some existing task allocation mechanisms choose users for task requirements,they do not know the categories that users are suitable for completing tasks,and do not consider that users have the ability to complete multi-category tasks,and once new tasks are pushed,the user information will be again provided The whole complexity of the mechanism is increased by processing.In order to solve these problems,this paper proposes a task assignment model which combines task demand feature extraction algorithm and user label classification method.The main contributions of this paper are as follows:(1)Combine the fuzzy clustering algorithm with the statistical topic model to extract the category keywords of the sensing task.(2)The data features are extracted through a multi-linear neural network,and multi-kernel learning is used to fuse and train the features to obtain a classifier,and the classifier is used to predict user type labels.(3)According to the category keywords of the tasks,combined with the spatial location information and user participation,the users who have the task category labels and meet the task requirements are selected to distribute tasks.