Abstract
Task assignment problem (TAP) has been extensively studied under different characteristics in recent years, because task assignment is a crucial step in distributed computing system where system performance and computing power usage can be improved by properly allocating tasks to processors. The target of the task assignment problem in distributed computing system is to assign program tasks to processors so as to minimize overall costs made of execution and communication costs and satisfy with various limited resources constraints simultaneously within a system. The task assignment problem with more than three processors is not only considered to be NP-hard, but also a nonlinear integer optimization problem, therefore it is difficult to find the exact solution for a large-scaled problem in an acceptable time. In this paper, we present an algorithm based on Simplified Swarm Optimization (SSO) to solve the problem, in the meantime, a local search optimizer based on the concept of entropy is embedded with SSO so as to better exploit local optimum. In addition, an earlier reported method called normalized penalty fitness function aimed to trade off the costs and the constraints is adopted in this paper. Afterwards, the results of the proposed simplified swarm optimization with entropy local search(SSO-E) is demonstrated by comparing with well-known Genetic method and other recently reported meta-heuristic algorithms such as Hybrid Particle Swarm Optimization (HPSO), Novel Global Harmony Search (NGHS) and Improved Differential Evolution (IDE). Finally, the experimental results show that that the proposed SSO-E produces superior quality of solutions among other compared methods on solving task assignment problem in distributed computing system.