The share of video contents over the Internet becomes larger as many OTT (Over-the-top) services begin to deliver the ordinary TV services, in addition to the conventional VOD and Internet video contents. In response to this trend, cloud-based video content delivery has recently been attracted much research attentions as a solution to reduce the traffic burden due to the heavy video contents over the Internet. In line with the recent researches on this topic, we formulated and solved the policy optimization problem in a cloud-based video distribution network and analyzed the policy sensitivity of delay performance. From the computational experiments performed for a hypothetical metropolitan area, we showed that the optimal content pre-fetching reduces significantly the expected delay and that the corresponding performance gain is most sensitive to the contents popularity and the prediction accuracy of the popularity.
展开▼