Research On Task Assignment And Incentive Mechanism For Spatial Crowdsourcing

With the rapid development of Internet commercial modes and mobile communication technologies,spatial crowdsourcing has received widespread attention as a distributed problem-solving mechanism.Unlike traditional crowdsourcing,spatial crowdsourcing focuses on tasks with location requirements such that workers must move to specific locations to complete tasks.Owing to the ability of leveraging crowd wisdom to solve complex tasks that the traditional crowdsourcing cannot accomplish,spatial crowdsourcing has attracted considerable interest from industry and academia.It is crucial to design efficient task assignment algorithms and incentive mechanisms in spatial crowdsourcing.One the one hand,the task-centric assignment algorithms can guarantee the system effectiveness and reliability.Although there are some achievements about task-centric assignment algorithms,yet these works have the following limitations: in the case of simple tasks,the issue of task expiration due to the hard time constraints and uncertain factors;in the case of complex tasks,the factors of task relevance and worker cooperation are ignored.On the other hand,the reasonable incentive mechanisms are of great significance to ensure the workers' enthusiasm for participation and optimization of task assignment when the workers' information and tasks' information are not completely public.Although some achievements about incentive mechanisms are made,yet those works do not consider the impact of task spatial coverage on the reliability of sensing platforms and the factor of workers' malicious competition.Consequently,in this dissertation,the task assignment problems and incentive mechanisms are studied by considering the factors of soft time window,tasks relevance and task spatial coverage,respectively.The main works are summarized as follows:(1)In the scenario of simple tasks,the task assignment problem of minimizing system cost is studied.First,the model is established by considering the initial cost requirements of workers and soft time window constraints of tasks,and the corresponding problem is proved to be an NP-hard problem.Second,as the computationally effective optimal solution for the problem cannot be found,and thus the task assignment algorithm is developed based on the discrete cuckoo search algorithm and the priorities of tasks.Finally,simulation results show that the developed algorithm outperform other algorithms.(2)In the scenario of complex tasks,the task assignment problem of maximizing task completion rate is studied.First,the model is formed by considering the workers cooperation and tasks relevance,and the corresponding problem is proved to be an NP-hard problem.Second,as the optimal solution cannot be derived within the polynomial time,the improved discrete cuckoo search algorithm is proposed to solve this problem.Finally,simulations results demonstrate that the proposed algorithm performs better than other algorithms.(3)In the scenario when the workers' information and tasks' information are not completely public,the incentive mechanism of maximizing platform revenue is studied.First,the model is developed by considering the workers' time constraints and task spatial coverage requirements,and the corresponding problem is proved to be an NP-hard problem.Second,the incentive mechanism based on the reverse auction and second-price auction is implemented,and the properties of individual rationality,computational efficiency,platform profitability as well as truthfulness are analyzed and confirmed for the incentive mechanism.Finally,simulations results verify that our mechanism achieves better performance than existing mechanisms.In this dissertation,the task allocation problems and incentive mechanism design of spatial crowdsourcing are studied by combining the factors of soft time window,task relevance and task space coverage.The related work can provide a reference for the system realization of spatial crowdsourcing.