A Novel Scheduling Algorithm based on Game Theory and Reinforcement Learning

In this paper we proposed a new dynamic scheduling algorithm for power scheduling problem. The algorithm is based on game theory and reinforcement learning approach. We compared the performance of our algorithm with that of online bin packing and MAB algorithm. We observed that our algorithm performs better than online bin packing when there is a variation in the deadlines. This is because our algorithm schedules the requests on the basis of their actions and the probability of missing the deadline and online bin packing algorithm schedules requests based on the sequence of requests as they arrive. We observed that our approach is more useful, when scheduling requests repeat themselves for long duration.