Chorus: Collision Resolution for Efficient Wireless Broadcast

Traditional wireless broadcast protocols rely heavily on the 802.11-based CSMA/CA model, which avoids interference and collision by conservatively scheduling transmissions. While CSMA/CA is amenable to multiple concurrent unicasts, it tends to degrade broadcast performance, especially when there are a large number of nodes and links are lossy. In this paper, we propose a new, drastically different protocol called Chorus that improves the efficiency and scalability of broadcast service with a MAC layer that allows packet collisions. Chorus is built upon the observation that packets carrying the same data can be effectively detected and decoded, even when they overlap in time and have comparable signal strength. It performs collision resolution using symbol-level iterative decoding, and then combines the resolved symbols to reconstruct the packet. This collision-tolerant mechanism significantly improves the transmission diversity and spatial reuse in wireless broadcast, providing an asymptotic broadcast delay that is proportional to the network radius. This advantage is exploited further by Chorus´s MAC-layer cognitive sensing and scheduling scheme. We evaluate Chorus with symbol-level simulation, and validate its network-level performance via ns-2, in comparison with a typical CSMA/CA broadcast protocol.