面向依赖关系约束的移动群智感知任务协作

Dependency constraint oriented task collaboration in mobile crowd sensing

现有移动群智感知中,大多研究将每个任务作为独立个体进行处理,对任务间约束关系缺乏研究,为此,提出了基于感知质量优先级的在线任务协作方法(online task collaboration method based on sensing quality priority,TCSP)。该方法首先使用贪婪算法计算感知质量优先级,对全部任务进行筛选以保证任务完成率;然后将选出任务中存在时间先后或执行逻辑前后关系的多个子任务构建为任务协作图,并将其协作过程建模为有约束的马尔可夫决策过程,通过强化学习算法求出最优协作策略。实验结果表明,与现有基线方法相比,所提出的任务协作方法能够减少依赖任务的平均完成时间,有效降低平台的平均感知成本。

In the existing mobile crowd sensing,most studies treat each task as an independent individual,and lack of research on the constraint relationship between tasks.In view of this,this paper proposed an online task collaboration method based on sensing quality priority(TCSP).Firstly,this method used greedy algorithm to calculate the sensing quality priority,screened all tasks to ensure the task completion rate.Then it constructed a task cooperation graph for multiple subtasks that had time sequence or execution logic relationship in the selected tasks and modelled cooperation process as a constrained Mar-kov decision process,obtained the optimal cooperation strategy through reinforcement learning algorithm.Experimental results verify that compared with the existing baseline methods,the proposed TCSP method can reduce the average completion time of dependent tasks and reduce the average sensing cost of the platform.