ECLIPSE: Error Correction based on Loss probability and data Importance for Plural Server Environment

Abstract

Video streaming technique is widely used in the Internet with the development of access networks. Although some techniques such as Forward Error Correction or Content Distribution Network are effective for video streaming services, they are not sufficient in the case that streams are suffered from heavy packet loss. For this problem, we propose a novel error correction method called “ECLIPSE: Error Correction based on Loss probability and data Importance for Plural Server Environment”. The ECLIPSE recovers lost packets with a parity stream which exploits multiple streams. In order to improve video quality, ECLIPSE gives priority to each data of multiple streams in terms of recovery from packet losses. It is designed for recovering important data with higher probability. The priority of each data is set based on two criteria. The one is the inherent importance of the data within one stream. The other is the loss probability of one stream which is compared with the loss probabilities of other streams. We also clarify the performance of the proposed method through theoretical analysis. Moreover, we focus on the creation method of redundant data. They are transferred from another server for multiple worse conditioned streams, which greatly affects the recovery capability. We propose the adaptive creation method for various network conditions.<br />Our simulation results show that the ECLIPSE can recover more important data with higher probability. Furthermore, it can also achieve almost the same restoration ratio among streams of recovery targets which have different loss probabilities