Tatiana Bokareva, Nirupama Bulusu and Sanjay Jha
Our performance study provides useful insights for the network designer --- such as which protocols (and design choices) scale control traffic well, improve data delivery or reduce overall energy consumption. We observe that despite the designers intentions to make these protocols self-configuring, they in fact rely on a number of statically configured parameters which are the cause of the reduction in performance. For example, the static preconfiguration of cell size in TTDD is one of the reasons why TTDD exhibits larger routing overhead than DD by 67.6% on average. Although GRAB produces approximately 93.6% smaller overhead than TTDD and 89.27% smaller than DD, because of statically configured credit amount, GRAB delivers on average 6 times more redundant data packets than TTDD and DD. We suggest that making these protocols truly self-learning can significantly improve their performance, and comment on how some of the parameters can be dynamically derived through measurements of network and event dynamics.
@article{Bokareva04a,
author = "Tatiana Bokareva and Nirupama Bulusu and Sanjay Jha",
title = "A Performance Comparison of Data Dissemination Protocols for Wireless Sensor Networks",
journal = "In Proceedings of IEEE Globecom Wireless Ad Hoc and Sensor Networks Workshop",
volume = "",
number = "",
year = "2004",
publisher = "IEEE Communications Society,"
month = "November",
pages = "",
keywords = "performance comparison; data dissemination; wireless sensor networks",
}