A Dynamic Task Allocation Algorithm Based on Weighted Velocity

Abstract

Volunteer computing is a way of supporting people around the world who provide free computer resources, to participate in scientific calculation or data analysis on the Internet. This provides an effective solution to solve the problems of large scale of basic scientific computing and more computing resources requirements. Task allocation is a very important part of volunteer computing. An effective algorithm can significantly improve computational efficiency. At present, most of the existing tasks are divided in term of the computer hardware conditions or the initial state of the computer in the volunteer computing. It seems that this have no obvious impact to calculating efficiency in a short time, but this task will be less flexible when idle resources of the volunteer computing becomes less or more. To make full use of idle computer resources, a dynamic task allocation algorithm (TAA) based on weighted velocity was proposed in this work. The research results showed that the weighted velocity as a parameter can be used to test the computing performance of a computer, dynamically manage task allocation as well.