Research On Tasks Management Technology For Collective Computing

With the popularity of the Internet of Things and Mobile Internet and the rapid development of Ubiquitous Computing and Cloud Computing technology,a new computing framework,Collective Computing,has emerged.Collective Computing provides a new solution architecture for widely complicated multi-source heterogeneous computing tasks,which makes full use of the pervasiveness,autonomy,heterogeneity of the collective,including humanity.It will be used to hand over a huge,complex task that cannot be completed by a single entity with effective task management mechanisms.Therefore,the extensive computing resources in the Internet can be fully utilized to effectively improve the efficiency of tasks execution.This thesis mainly divides the tasks management technology in Collective Computation into two parts,tasks decomposition and tasks allocation.At first,this thesis proposed an extended task-tree model to decompose tasks by dividing the association between tasks into three types: “Serial”,“Concurrent”,“Or”.At the same time,using a series of algorithms for the trim and reorganization of the task-tree can ensure improving task parallelism while avoiding excessive decomposition.That is,under the premise of minimizing completion time,refraining the number of tasks are too much which will increase the pressure of the tasks allocation process and the demand of computing resources.Besides,there's a task dependency graph constructing by the task-tree and the associations between the nodes obtained by the above task decomposition algorithms,and put forward the algorithm(LLRF-GA)using the task dependency graph to achieve tasks allocation.The algorithm firstly obtains the priority of each task according to the laxity ratio calculation formula proposed in this paper to get the task allocation order,and then use the two-stage genetic algorithm to optimize the initial population trying to find a near global optimal task allocation method.Finally,this thesis develops a new Collective Computing simulation system to simulate the process of tasks decomposition and tasks allocation.At the same time,the simulation system is used to analyze and evaluate the performance of a series of task decomposition and allocation algorithms proposed in this thesis.The experimental results also prove the efficiency and superiority of the proposed algorithm.