Time-sensitive And Quality-aware Incentive Mechanism For Social Welfare Maximization In Mobile Crowdsensing

Mobile Crowd-Sensing(MCS)is a well-known paradigm that exploits users with smart devices to provide complex sensing tasks.MCS has several advantages,such as scalability,low cost and good spatiotemporal coverage.The users accept to perform the sensing task,and the final sensed data is transmitted to the platform server.These smart device users consume time,energy,and computing and communication resources during the sensing process.In addition,they risk having their private information,such as their trajectory,unique interest or identity,compromised.Therefore,monetary or entertainment rewards are required to encourage users to participate in the sensing tasks.In this thesis,we start the research by studying the preliminaries of MCS and evaluating the state-of-art incentive mechanism in the current literature.First,time sensitivity and sensing quality awareness are used as metrics to evaluate the previous papers on incentive mechanisms.Then,we focus on literature that studied incentive mechanisms to maximize social welfare in MCS.However,it was observed that the literature had some unrealistic assumptions,such as assuming that each user can be assigned to a single sensing task,had ample time to complete the tasks,and could perform the allocated task with high-quality sensing.Based on the information gained from the literature,we propose Time-sensitive and Quality-Aware incentive Mechanism(TSQM)to solve the social welfare optimization problem under time,sensing quality and budget constraints by jointly allocating appropriate sensing tasks to a set of users and paying them sufficiently.We show that the optimization problem consists of two stages: task allocation and payment.The task allocation is formulated as a multiple sum of subsets problem,proven to be Non-deterministic Polynomial(NP)-complete and solved using Genetic algorithm.On the other hand,a payment mechanism is introduced to determine the appropriate reward given to each selected user to maximize social welfare.We determine the efficiency of TSQM compared to three state-ofthe-art approaches from the literature by evaluating their average performances,social welfare,and task allocation ratio based on varying the number of users,tasks and the size of the tasks time window in each iteration.General simulation results confirm the superiority of TSQM in comparison to the other approaches.In addition,we study the users’ performance based on their utility(i.e.,profit).For this purpose,we propose an incentive mechanism known as Worker Multitask Allocation-Genetic Algorithm(WMTA).This algorithm aims to allocate many suitable tasks to the users to maximize utility.Extensive simulation based on synthetic data set is used to determine the performance of WMTA compared to other baseline approaches.