Research Of Workflow Assignment Algorithm Base On Mobile Crowdsensing

Workflow scheduling problem has been a hot topic in the field of workflow and research.An efficient workflow scheduling algorithm can greatly improve the service quality of applications.The role of workflow scheduling algorithm is to allocate a variety of computing resources to different workflow tasks reasonably,as to meet customer's service quality requirements.In recent years,with the rapid development of mobile internet,the number of mobile users is increasing.In this context,a mobile groupware awareness model has been developed.It mainly uses the cognitive,computing and wireless communication functions of mobile intelligent devices to build a human-centric perceptual network,and utilizes the shared resources in mobile smart devices to accomplish large-scale and complex applications.However,due to the limitation of various execution resources in mobile devices,all the tasks in the mobile workflow can't be completed by a single individual.At present,most studies were uploaded tasks to the cloud for execution,and such cloud-based methods require a large cost.Therefore,this paper proposes a workflow scheduling method based on mobile groupware awareness,which not only needs to consider the completion time of tasks in mobile group environment,but also needs to save the resource overhead of mobile users.At the same time,workflow scheduling based on group-wise perception needs to consider the dynamics,distribution,heterogeneity and autonomy of mobile computing.These features lead to the traditional workflow methods and technologies can't effectively deal with such issues under these circumstances.In view of the dynamics of mobile devices under mobile computing,this paper regards the movement of swarm intelligence as a research direction in the field of workflow.In this paper,we will focus on the task allocation in the process of mobile swarm intelligence perception.This paper presents a mobile group intelligent network model that describes the user scheduling process.Through the analysis and specification of the problem,it proves that the problem is NP-hard,and introduces the Q-Learning thought in intelligent control into the workflow task distribution system.Reduce the application's maximum completion time allocated between dynamically-aware users.Aiming at heterogeneity and multi-objective problem of mobile computing,a user heterogeneous model is first proposed and a multi-objective optimization model is proposed by studying Pareto optimal solution.Then a multi-objective optimization model greedy iterative method.The algorithm learns and optimizes from the macro level through every exploration of reinforcement learning.At the micro level,the greedy algorithm chooses the local optimal solution for each iteration,which enhances the performance of the algorithm and accelerates the convergence speed.Finally,this article selects two commonly used open source mobile device connection trajectory to verify the experiment.Experimental results show that,compared with the other three algorithms,the proposed algorithm can not only reduce the time and energy consumption overhead,but also converge faster and improve the perceived efficiency.The task allocation algorithm studied in this paper can be universally applicable to similar problems,has a high degree of universality,can be extended to similar scenarios and models,and can be used as an intelligent mobile workflow agent scheduling problem.