I am also interested in applying artificial intelligence and machine learning methods in job scheduling for GRID and CLOUD computing. That's why I need a way for simulating GRID for experiments. I will read carefully the paper You shared. If I have questions, I will ask. Thank You sincerely, Mohammad!
I think, first of all, you should provide a model that helps you to map your scheduling. For AI scheduling, one of the important step is to know more about AI method you are going to use. For example, it is important in GA to know Mutation, Crossover and fitness functions. Second, you apply the parameters in your work. E.g., Energy consumption, makespan, execution time, processor speeds, bandwidth allotting, the cost which are considered for each tasks or jobs, and the status of your jobs (independent, dependency) etc. If you interested more, you are able to check the attached file that explains an AI method for scheduling and resource allocation in Grid. it belongs to reputation-based approaches applied in economic grid. if you have question feel free to ask. good luck.