Maximum Capacity of Heterogeneous P2P Multimedia Streaming Network

Online P2P multimedia streaming systems made great success in the past few years. One of the important performance measurements of the P2P streaming systems is capacity. It is defined as the largest rate that can be concurrently achieved by all the receivers in a multicast session. The capacity depends on system configurations and constraints, including bandwidth, degree bound of each node, coding schemes, etc. Usually the approximate algorithm is used to approach the maximum capacity under these constrains, since the capacity computation has been proved to be NP-complete. Previous theoretic studies decompose the mesh-based network into multiple trees and approximately derive the maximum capacity of these trees. In this paper, we propose an alternative heuristic algorithm to find the maximum capacity by visualizing the architecture of a heterogeneous P2P network as a ring instead of trees. Comparing with the near optimal tree-based algorithm proposed in [1], our algorithm is able to derive larger capacity in 98.47% tests, while requiring less outgoing degrees of nodes in 98.06% tests. Its complexity is only O(n^2), where n is the number of nodes in the system.