Skip to Main Content
In this paper, we propose an admission control method based on model and prediction for distributed service systems. First, we use partly observable Markov decision process (POMDP) to model the service system. Next, we put the service allocation policy into the model parameters and use randomized policy as admission control policies to optimize system performance. The target of optimization is to maximize the system benefits. Based on the POMDP model, we propose an observation-based policy gradient algorithm to solve the optimal policy. We use HMM-based method to detect and predict change of the system, then updating the system model and admission policy with dynamic adaptive method. Experiment result shows compare with the best effort service policy our optimal policy has a better system performance.