Research on Fuzzy Reinforcement Learning Algorithm for Agents in Grids

How to improve the efficiency and performance of job scheduling in grid computing is one of the most important and challenging techniques. This paper tries to give out a novel grid job scheduling model based on agent technology. To make full use of intelligence and adaptability of the agents, dynamic fuzzy knowledge-base and corresponding fuzzy reinforcement learning algorithm are proposed for the job scheduling agents. The model and algorithm can largely meet the needs of intelligence, flexibility, scalability and optimization for grid job scheduling. Simulation experiments show that the proposed reinforcement learning algorithm for agents based on dynamic fuzzy knowledge-base works better compared with other similar learning algorithm.