Synthesized Trees
Home | Publications | Download | Instructions
Synthesized Trees | Efficiency Study | Anatomy Study
Using the generate facility of mwalk, we have collected paths for almost 2000 receivers. Using a local source at UCSB, we traced paths from a semi-random set of over 20,000 receivers in the Internet. From our studies of bandwidth efficiency and topology, we feel that this synthesized receiver distribution conforms to many properties of realistic inter-domain multicast groups.

We are offering subsets of this synthesized topology in the hope that on-going research efforts in multicast reliability, congestion control and routing protocols can benefit from "realistic" group distributions. We have taken a number of random samples of different sizes from the complete topology. The files are simple ASCII. Each node of the tree is assigned a unique id and each line of the file represents a single node with the following format:

    node-id [child-id ...]
        
The first number is the node's own id. Any subsequent numbers are the ids of the node's downstream neighbors (children). For example:
        1 2 3
        2 4
        3 5 6
        4
        5 7 8
        6
        7
        8
	
would represent the tree,
              1
             / \
            2   3
           /   / \
          4   5   6
             / \
            7   8 
	

Currently, the ids are simply integers that have no relation from one dataset to another. We decided not to release the actual traced IP addresses to provide some level of anonymity. If you are interested in the complete data or correlated datasets, please contact us at robertc@cs.ucsb.edu.

contact us at robertc@cs.ucsb.edu updated 05.01.01