Detecting Malicious Peers in Overlay Multicast Streaming

Overlay multicast streaming is built out of loosely coupled end-hosts (peers) that contribute resources to stream media to other peers. Peers, however, can be malicious. They may intentionally wish to disrupt the multicast service or cause confusions to other peers. We propose two new schemes to detect malicious peers in overlay multicast streaming. These schemes compute a level of trust for each peer in the network. Peers with a trust value below a threshold are considered to be malicious. Results from our simulations indicate that the proposed schemes can detect malicious peers with medium to high accuracy, depending on cheating patterns and malicious peer percentages.