PeerCast: Improving link layer multicast through cooperative relaying

• Main Authors: XIONG, Jie, CHOUDHURY, Romit Roy
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2011
• Subjects: Multicast protocols; Radiocommunication; MobiTV; PeerCast; Access point; Cooperative relaying; Link layer multicast; Multimedia classrooms; Single packet transmission; Web telecast; Wireless channel; Wireless multicast application; Wireless multicast protocol; Computer Sciences; Software Engineering
• Online Access: https://ink.library.smu.edu.sg/sis_research/2797; https://ink.library.smu.edu.sg/context/sis_research/article/3797/viewcontent/peercast_infocom11.pdf
• Institution: Singapore Management University

Wireless multicast applications, such as MobiTV, web telecast, and multimedia classrooms, are gaining rapid popularity. The broadcast nature of the wireless channel is amenable to such multicasts because a single packet transmission can be received by all clients. Unfortunately, the rate of this transmission is bottlenecked by data rate of the weakest client, degrading system performance. Attempts to increase the data rate results in lower reliability and higher unfairness. This paper presents PeerCast, a wireless multicast protocol that engages clients in cooperative relaying. The main idea is simple. Instead of multicasting at the bottleneck rate, the access point transmits at a high rate and suitably chooses a few stronger clients to relay the packet to the weaker ones. Multiple transmissions of the same packet, each at higher rate, can achieve better throughput than one transmission at the low, bottleneck rate. We propose a new simultaneous reply-back scheme for clients and detect the power level to estimate the AP's transmission strategy. PeerCast translates these ideas into a functional system using off-the-shelf hardware. Performance evaluation on a 9 node testbed demonstrates consistent throughput and reliability improvements over 802.11. Simulations in QualNet indicate similar trends in large-scale networks.