Distributed Hierarchical Multicast Tree Algorithms for Application Layer Mesh Networks

Weijia JIA  Wanqing TU  Jie WU  

Publication
IEICE TRANSACTIONS on Information and Systems   Vol.E89-D   No.2   pp.654-662
Publication Date: 2006/02/01
Online ISSN: 1745-1361
DOI: 10.1093/ietisy/e89-d.2.654
Print ISSN: 0916-8532
Type of Manuscript: Special Section PAPER (Special Section on Parallel/Distributed Computing and Networking)
Category: Parallel/Distributed Algorithms
Keyword: 
application layer multicast,  distributed computing,  mesh network,  short multicast delay,  low resource consumption,  

Full Text: PDF(606.9KB)>>
Buy this Article




Summary: 
This paper proposes a set of novel distributed algorithms on m-D mesh overlay configurations for short delay and low resource consumption application layer multicast. In contrast to previous approaches, our application layer multicast adopts two-layer tree architecture and the novelty and contribution are: (1) cluster formation algorithm assigns the closest group members into the same cluster that greatly decreases the multicast delay and resource consumption caused by the message transmission among the members with long distances; (2) optimal core selection algorithm seeks the cluster member who has the minimum sum of static delay distances to other cluster members as the optimal cores (i.e. cluster cores) that guarantees the short multicast delay; (3) weighted shortest path tree generation algorithm constructs a shortest path tree rooted at the optimal core for each cluster. The shortest path tree utilizes the minimum sum of links that are on the shortest paths among the cluster members; and (4) distributed multicast routing algorithm directs the multicast messages to be efficiently distributed along the two-layer multicast architecture in parallel without a global control. The extended simulation results indicate that the application layer multicast constructed by our algorithms is efficient in terms of short multicast delay and low network resource consumption as compared with other well-known existing multicast solutions.