A Resilient Video Streaming System Based on Location-Aware Overlapped Cluster Trees

Tomoki MOTOHASHI  Akihiro FUJIMOTO  Yusuke HIROTA  Hideki TODE  Koso MURAKAMI  

Publication
IEICE TRANSACTIONS on Communications   Vol.E96-B   No.11   pp.2865-2874
Publication Date: 2013/11/01
Online ISSN: 1745-1345
DOI: 10.1587/transcom.E96.B.2865
Print ISSN: 0916-8516
Type of Manuscript: PAPER
Category: Network
Keyword: 
P2P,  live streaming,  application level multicast,  ALM,  

Full Text: PDF>>
Buy this Article




Summary: 
For real-time video streaming, tree-based Application Level Multicasts (ALMs) are effective with respect to transmission delay and jitter. In particular, multiple-tree ALMs can alleviate the inefficient use of upload bandwidth among the nodes. However, most conventional multiple-tree ALMs are constructed using a Distributed Hash Table (DHT). This causes considerable delay and consumes substantial network resources because the DHT, generally, does not take distances in the IP network into account. In addition, the network constructed by a DHT has poor churn resilience because the network needs to reconstruct all the substreams of the tree network. In this paper, we propose a construction method involving overlapped cluster trees for delivering streamed data that are churn resilient. In addition, these overlapped cluster trees can decrease both the delay and the consumption of network resources because the node-connecting process takes IP network distances into account. In the proposed method, clusters are divided or merged using their numbers of members to optimize cluster size. We evaluated the performance of the proposed method via extensive computer simulations. The results show that the proposed method is more effective than conventional multiple-tree ALMs.