An algorithm for cost-delay balanced trees and its application to wide area network multicast

Abstract

We present a distributed algorithm for creation of a multicast communication tree based on data-flow discovery. Each receiver chooses its preferable Quality of Service (QoS), namely cheaper cost or shorter delay. For shorter delay QoS, the algorithm uses the shortest path from the receiver to the sender. If a receiver prefers cheaper cost QoS, then the algorithm tries to find an already existing data flow within the receiver’s a~ea. Using a parameter we can control the size of the search area so that we can produce a cost/delay optimized communication tree. First we present the fiat network version of the algorithm where we use local broadcast to find an already existing data flow. The simulation shows that we can save up to 30% of the tree cost compared to the Shortest Path Tree (SPT). Secondly we present the hierarchical version of our algorithm. We use a Multicast Registration Server tree to search for an existing data flow and to reduce the amount of broadcasted messages.