Splash: Fast Data Dissemination with Constructive Interference in Wireless Sensor Networks
Authors:
Manjunath Doddavenkatappa, Mun Choon Chan, and Ben Leong, National University of Singapore
Abstract:
It is well-known that the time taken for disseminating a large data object over a wireless sensor network is dominated by the overhead of resolving the contention for the underlying wireless channel. In this paper, we propose a new dissemination protocol called Splash, that eliminates the need for contention resolution by exploiting constructive interference and channel diversity to effectively create fast and parallel pipelines over multiple paths that cover all the nodes in a network. We call this tree pipelining. In order to ensure high reliability, Splash also incorporates several techniques, including exploiting transmission density diversity, opportunistic overhearing, channel-cycling and XOR coding. Our evaluation results on two large-scale testbeds show that Splash is more than an order of magnitude faster than state-of-the-art dissemination protocols and achieves a reduction in data dissemination time by a factor of more than 20 compared to DelugeT2.
USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.
BibTeX
@inproceedings {180310,
author = {Manjunath Doddavenkatappa and Mun Choon Chan and Ben Leong},
title = {Splash: Fast Data Dissemination with Constructive Interference in Wireless Sensor Networks},
booktitle = {10th USENIX Symposium on Networked Systems Design and Implementation (NSDI 13)},
year = {2013},
isbn = {978-1-931971-00-3},
address = {Lombard, IL},
pages = {269--282},
url = {https://www.usenix.org/conference/nsdi13/technical-sessions/presentation/doddavenkatappa},
publisher = {USENIX Association},
month = apr
}
This paper presents Splash, a dissemination protocol for large objects in wireless sensor networks. Dissemination from one node to many is an important service in wireless sensor networks, used for code updates, for example. Splash combines recent advances in constructive interference broadcast and multiple-channel pipelining to eliminate contention overhead among nodes. Constructive interference is a broadcast technique that is emerging as the currently best-known approach for low-power wireless communication.
Splash's constructive interference is based on Glossy, which established that concurrent transmissions by nodes synchronized to within 0.5 microseconds can significantly improve reception rates by neighbors. While Glossy is designed to flood a single packet at atime, Splash disseminates a stream of packets that require 100% reliability. To this end, Splash extends a multi-channel pipelining technique presented in PIP to work on a tree rooted at the source. In addition to these two basic techniques, Splash has a number ofimportant mechanisms that contribute to is good performance.
The algorithm works in rounds based on the depth of nodes in the tree. Nodes at the same level are synchronized by the reception of the previous round, and transmit at the same time in practice, due to deterministic delay in the relays. Successive levels of the treetransmit in alternating channels, to improve pipelining. If a node fails to receive a packet in a round, it immediately switches to receive mode, to try to recover the packet from its siblings, who are about to relay the packet to the next level. Every packet istransmitted in two rounds: in the first round only non-leaf nodes retransmit, and in the second round all nodes retransmit. This increases density diversity, as the effectiveness of constructive interference may not vary monotonically with density. Splash uses athird round to recover the (likely few) missing packets at different nodes. In this round each node broadcasts the packets it has XOR'ed with 19 other packets, as they found this greatly increases the chance of recovery. Finally, nodes locally ask their neighbors forindividual missing packets, a process that converges quickly.
The paper evaluates a full implementation of Splash on two testbeds, and presents very impressive results, achieving over an order of magnitude faster dissemination than previous approaches.
connect with us