Flash Flooding



Research Publications Misc
Overview The Flash protocol enables multiple nodes to transmit concurrently in order to eliminate neighborhood contention and reduce the latency of network floods in wireless networks. Flash relies on the capture effect to ensure that packets are received during concurrent transmissions, exploiting the fact that all nodes in a flood transmit the same message so that each node can receive the message from any of its neighbors. A key issue in the design of Flash is to manage transmission concurrency: too much results in packet collisions while too little leads to neighborhood contention. Flash proposes two flooding- specific mechanisms to manage concurrency. The first, used in Flash-II for both high- and low-duty networks, is to transmit the flooding packet twice, with CCA and MAC delay in between. This allows the network to recover from floods that stop prematurely due to high concurrency. This protocol can approach the theoretical lower bound on flooding latency and sacrifices flooding throughput for reliability. The second mechanism, used in Flash-III for low-duty cycle networks only, is to use a large IPS and a very small CCA. The effect of this combination is to allow some but not too much concurrency and operate on the probability that the CCA of a node overlaps with the IPS of all nodes currently transmitting. This protocol approaches the theoretical lower bound on flooding latency and does not compromise reliability or power consumption. Flash is the first network flooding protocol for wireless networks that explicitly exploits the capture effect to optimize for latency. The implementation of Flash only involves changes to tens of lines of code with respect to the standard CC2420 radio stack of TinyOS, which will allow this protocol to have an immediate and practical impact.
Publications

Jiakang Lu, and Kamin Whitehouse. Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks. The 28th Conference on Computer Communications (INFOCOM 2009).  Rio de Janeiro, Brazil, Apr. 2009.
(19.6% acceptance rate)

Jiakang Lu, and Kamin Whitehouse. "Poster Abstract: Exploiting the capture effect for low-latency flooding in wireless sensor networks". The 6th ACM Conference on Embedded Networked Sensor Systems (SenSys).  Raleigh, NC, Nov. 2008.



Kamin Whitehouse
Computer Science Department
The University of Virginia
217 Olsson Hall
Charlottesville, Virginia 94720