摘要

In this paper, we study admission control problems of P2P-based Media Delivery Network (P2P-based MDN). We propose a specific partially observable Markov Decision Process (POMDP) model for PEP-based MDN. Based on this model, the observation-based randomized policy for admission control is provided, which optimizes system's performance by applying policy-gradient algorithm. Observation-based policy can enhance the real-time performance of the system, and policy optimization can improve the accuracy of controller's judgments based on partial information. Novel source selection policy and bandwidth allocation policy are designed to reduce service delay and provide high-quality service. A numerical example is provided to illustrate the effectiveness of our methods.

全文