Research Based On Divisible Load Theory In Computing Load

With the development of the times, mobile phones, computers, tablet to computes in our lives more and more popular. More and more available computing resources around us, but for some complex applications, for example, a large number of experimental data processing, image processing matrix calculation, calculation on a single device may take longer, affecting people’s experience, in this case, the calculation unloading emerged. However, for the calculation unloading, not be beneficial in all cases, for example, when the communication costs of a task to be higher than the cost of unloading, this time not conducive to calculate unloaded, only when the cost of unloading less than the cost of communications, the loading can help us.Computing unloading can solve the problem of lack of computing resources on the local computing device, for example, memory, energy, lack of storage space and other issues. The divisible applications are a class of tasks that permit the partition of the processing load into several smaller fractions to be processed independently, at last, each part of the last task can be combination. This property make them suitable for being processed in opportunistic computing environments. In our lives, we may encounter a situation where a group of friends to travel, want the text converted into local photo mother tongue, it is possible to use OCR recognition, This paper addresses the problem of scheduling arbitrarily divisible load applications in opportunistic computing environments when OCR as divisible applications, increase the usefulness of this article.The main works are as below:(1) Proposed the mathematical model for scheduling arbitrarily divisible load applications in opportunistic computing environments.(2) The scheduling problem of OCR application expressed as a linear function, we proposed a set of linear equations. On this basis, we propose to solve the model of how to divide and allocate divisible load applications, in order to minimize completion time and energy consumption.ect.(3) Proposed performance evaluation program, to build a bench, simulation and practical experiments. The results show that our model and the program is effective and practical.