Alberto Cerpa, Jennifer Wong, Louane Kuang, Miodrag Potkonjak, Deborah Estrin
Recently, several landmark wireless sensor network deployment studies clearly demonstrated a great discrepancy between experimentally observed communication properties and properties produced by widely used simulation models. Our goal is to accurately model the relationship between the communication and location properties of experimentally observed networks. Our first goal is to provide sound foundations for conclusions drawn from these studies by extracting the relationship between pairs of location (e.g distance) and communication properties (e.g. reception rate) using nonparametric statistical techniques and by calculating intervals of confidence for all claims. The objective is to determine not only the most likely value of one feature for an alternate given feature value, but also to establish a complete characterization of the relationship by providing a probability density function (PDF). The PDF provides the likelihood that any particular value of one feature is associated with a given value of another feature. Furthermore, we study not only individual links properties, but also their correlation with respect to common transmitters and receivers and their geometrical location. The set of studied properties is comprehensive enough to capture all properties that are essential for consideration during algorithm and protocol development. The second and main objective is to develop a series of wireless network generators which produce networks of an arbitrary size and under arbitrary deployment rules with realistic communication properties. For this task we use a generalized rejection algorithm and an iterative improvement-based optimization procedure to generate instances of the network that are statistically similar to empirically observed networks. The key technical highlight of the research is a perturbation-based procedure that enables evaluation of the ability of a generated wireless network of arbitrary size to capture and generalize the properties of smaller deployed networks. We evaluate the accuracy of the conclusions drawn using the proposed model and therefore comprehensiveness of the considered properties on a set of standard communication tasks, such as connectivity maintenance and routing. The third goal is to use the developed models and generators to identify new aspects of standard communication tasks, the needs for new design and evaluation methodologies, new metrics, new theoretical development, and new tasks themselves. We address one of these tasks, the radio assignment problem, using algorithmic and evaluation techniques.
Alberto Cerpa, Jennifer Wong, Louane Kuang, Miodrag Potkonjak, Deborah Estrin, "Statistical Model of Lossy Links in Wireless Sensor Networks," Proceedings of the Fourth ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN 2005), pp. 81--88, ACM/IEEE, Los Angeles, CA, USA, July, 2005.
@Conference{Cerpa05a,
author = "Alberto Cerpa and Jennifer Wong and Louane Kuang and
Miodrag Potkonjak and Deborah Estrin",
title = "Statistical Model of Lossy Links in Wireless Sensor
Networks",
booktitle = "Proceedings of the Fourth ACM/IEEE International
Conference on Information Processing in Sensor Networks
(IPSN 2005)",
year = "2005",
pages = "81--88",
address = "Los Angeles, CA, USA",
month = jul,
publisher = "ACM/IEEE",
URL = "http://www.andes.ucmerced.edu/papers/Cerpa05a.pdf",
accept = "20",
cited = "266",
}