Research On Task Allocation In Mobile Swarm Intelligence Perception Network

With the popularity of mobile smart devices,mobile crowd sensing emerges as a new sensing network.Task assignment has always been one of the key in the research of mobile crowd sensing.There are various types of tasks in mobile crowd sensing.such as offline tasks which are previously known,online tasks with uncertain delivery time,and spatial tasks with specific location.At present,there are few studies on the assignment of mixed spatial tasks.The contents of this study are as follows:For the assignment of mixed spatial tasks,this paper designs three assignment mechanisms,which are CostFirst&&AM-ONLT,RatioFirst&&AM-ONLT and GGA-TA&&AM-ONLT respectively.These three task allocation mechanisms adopt three different off-line spatial task allocation algorithms and the same online spatial task allocation algorithm.For off-line spatial tasks,this paper designs three different task allocation algorithms,including CostFirst algorithm based on greedy strategy,which gives priority to users with low cost.RatioFirst algorithm based on greedy strategy,which gives priority to users with high ratio of sensing quality to cost and GGA-TA algorithm based on the genetic algorithm and greedy strategy,which takes into account users' sensing quality,sensing cost and execution ability comprehensively.Simulation experiments have verified that,in the problem of off-line spatial task allocation,the distribution obtained by GGA-TA algorithm is lower in the sensing cost and distance user traveled than the results of CostFirst and RatioFirst algorithms,and the execution time is much higher than that of CostFirst and RatioFirst algorithms.This paper designs an AM-ONLT algorithm for online spatial tasks with uncertain delivery time.Based on the greedy strategy,the algorithm assigns the online tasks to the user who performs the nearest task under the condition that the task sensing time requirement is satisfied.In addition.aiming at the randomness of user sensing quality,the user's sensing quality is updated according to the user's historical implementation and current task execution.Simulation results show that the distribution of GGA-TA&&AM-ONLT is lower than the allocation result of CostFirst&&AM-ONLT and RatioFirst&&AM-ONLT in the aspect of sensing cost and distance traveled by users.and the execution time is much higher than that of CostFirst&&AM-ONLT and RatioFirst&&AM-ONLT.The distribution of RatioFirst&&AM-ONLT is lower than CostFirst&&AM-ONLT,higher than GGA-TA&&AM-ON LT in the aspect of sensing cost and distance traveled by users,and its execution time is much lower than that of GGA-TA&&AM-ONLT.