Paper Info
Title
Integrated prefetching and caching for adaptive video streaming over HTTP: an online approach
Abstract
We present an integrated prefetching and caching proxy, termed iPac, for HTTP-based adaptive video streaming services like Netflix and YouTube. The challenge we address is maximizing the byte-hit ratio for proxies through prefetching in the context of the limited bandwidth between proxies and content servers. The problem is NP-hard, and the best approximation ratio that any optimal offline algorithm can achieve is 1-e--1 ≈ 0.63. Considering that offline algorithms cannot be applied to real-time applications with stringent time constraints, we propose a novel 0.5-competitive online prefetching algorithm which, to the best of our knowledge, has the best lower bound so far. We evaluate the performance of iPac by deploying it over the Amazon EC2 cloud accepting user requests from the video clients deployed on the PlanetLab based on a real trace of user requests for YouTube videos. Our experimental results demonstrate that iPac can significantly improve the performance in terms of byte-hit ratio (up to 84%) and video rates (up to 34%), compared with the state-of-the-art approaches. The proposed iPac is compatible with existing HTTP-based adaptive streaming implementations without requiring any modification to existing content servers and video clients.
Year
DOI
Venue
2015
10.1145/2713168.2713181
MMSys
Keywords
Field
DocType
prefetching,algorithms,miscellaneous,online algorithm,online computation,adaptive streaming,caching,performance,performance attributes
Online algorithm,PlanetLab,Upper and lower bounds,Computer science,Video streaming,Server,Computer network,Implementation,Real-time computing,Bandwidth (signal processing),Cloud computing
Conference
Citations 
PageRank 
References 
13
0.68
20
Authors
4
Name
Order
Citations
PageRank
Ke Liang1132.71
Jia Hao2754.74
Roger Zimmermann31735147.98
David K. Y. Yau4502.88