Scheduling Workflow Tasks in Complex Networks on Mobile Clouds

Mobile cloud computing technology enables users to cooperate with each other so that their mobile devices' resources can be integrated together to afford complicated application, such as machine auto-translation and image processing, which can always be represented as Directed Acyclic Graphs. On mobile clouds, due to the social relationships of mobile devices' owners, networks on mobile clouds are always complex and consist of multiple sub-networks intersecting at several joint devices. In these complex networks, joint devices can communicate with all devices of their related multiple sub-networks, whereas the other devices can only contact devices in the same sub-nebvork. In this paper, we study the workflow task scheduling problem in the aforementioned complex networks on mobile clouds, and formulate it as Integer Programming problem that is generally NP-Hard. Due to the constraint of each device's power capacity, the dependence among tasks in a workflow and the network complexity, it is too difficult to achieve feasible scheduling solutions for the considered problem. To solve this problem, we propose an Improved Round-Robin scheduling algorithm (IRR) and an Improved Greedy scheduling algorithm (IG) with the consideration of workflow tasks' features. Experimental results show that our proposed IRR and IG guarantee to achieve feasible solutions, whereas the General Round-Robin scheduling algorithm (GRR) does with the probability of 2.3%. Moreover, experimental results also indicate that IG outperforms ERR.