Quality Of Information Oriented Research On Localized Collaboration Methods For Mobile Crowd Sensing With Heterogeneous Devices

In mobile crowd sensing,the mobile devices are integrated with multiple sensing interfaces and communication ports.The mobile devices are widely distributed and the crowd of mobile devices can collect and share variety kinds of sensing data.However,the sensing abilities,communications ports and the mobility of the crowd sensing devices are heterogeneous.The heterogeneity is contradictory with the Quality of collected sensing Information(QoI).To improve the quality of sensing information,the current study focused on the mobile nodes recruiting,selecting and sensing task allocating,lacking of optimizing the execution of sensing task.This article proposed sensing task method based on utility and sensing task offloading with task split method among the distributed heterogeneous mobile devices which could collect data collaboratively.In the end,it has a higher recognition performance in the ratio of sensing data and the ration of sensing task finished.The research contents of this thesis are shown as follows:A mobile crowd sensing system structure is proposed.The mobile crowd sensing system consists of sensing task publishers,a sensing data service center,wireless networks and a set of option mobile nodes which move randomly in the urban area.The sensing task publisher is a person or entity which published sensing tasks.The sensing data server is responsible for sending sensing tasks to mobile nodes and storing sensing data.The mobile nodes can exchange information of mobile trace and sensing utilities with other via wireless network and upload sensing data to sensing data service center.A Markov chain based on mobility model is established.The movement and distribution of sensing nodes are random.In order to overcome these drawbacks of collecting sensing data,a mobility model is established based on the discrete Markov chain.A utility function is constructed.The utilities of sensing nodes are used to describe the sensing abilities of the sensing nodes in the following period time.The arrival time distribution and the residence time distribution of mobile sensing nodes to a certain sensing region are modeled.A sensing task offloading algorithm based utility and a sensing task offloading algorithm with task split are proposed in this thesis.