Research On Task And Resource Allocation Mechanism In Crowdsening System

With the rapid development of wireless communication devices and new technologies,the number of smart devices has increased exponentially,which has promoted the rise of network applications based on mobile terminals.As one of the important applications of this kind of emerging network technology,Crowdsensing came into being.Crowdsensing aims to reduce the time and cost of large-scale data collection in real time by acquiring various types of information in a wider sensing area through smart devices attached to people such as smart phones and wearable devices.Crowdsensing system includes a cloud data platform and several users.The cloud data platform has a large number of perceived tasks that need to be executed and a large amount of idle computing resources.Therefore,the platform needs to assign perceived tasks and computing resources to users.It can be seen that the assignment of tasks and resources is two extremely important issues in the crowdsensing system.In order to solve these problems,this thesis has done the following work based on the existing research results:(1)The existing incentive mechanism requires users to estimate the cost and frequently bid,which reduces the willingness of users to participate in the task.In order to solve this problem,this thesis designs an offline completely-competitive-equilibrium-based crowd-sensing task allocation mechanism,which uses game theory to model the research problems.(2)Since the offline crowdsensing task allocation mechanism is not flexible,it cannot provide sufficient incentives for users.This thesis proposes an online complete-competitive-equilibrium crowdsensing task allocation mechanism based on the offline complete-competitive-equilibrium crowdsensing task allocation mechanism.(3)Aiming at the resource allocation problem in crowdsensing,this thesis takes the platform revenue maximization as the optimization goal,and designs an approximate opti-mal resource allocation mechanism to encourage the platform to release resources to third-party resource requesters.The mechanism can simultaneously support three types of re-source requirements that may exist in the users,and achieve more flexible allocation of cloud computing resources.In addition,we theoretically prove that the approximate ratio of the designed mechanism is max1≤m≤M(cm+1)cm/(Cm+1)Cm-Cmcm,where cm is the number of virtual machines of type m.