Dozer: Ultra-Low Power Data Gathering in Sensor Networks
Nicolas Burri
Computer Engineering and
Networks Laboratory
ETH Zurich
8092 Zurich, Switzerland
nburri@tik.ee.ethz.ch
Pascal von Rickenbach
Computer Engineering and
Networks Laboratory
ETH Zurich
8092 Zurich, Switzerland
pascalv@tik.ee.ethz.ch
Roger Wattenhofer
Computer Engineering and
Networks Laboratory
ETH Zurich
8092 Zurich, Switzerland
wattenhofer@tik.ee.ethz.ch
ABSTRACT
Environmental monitoring is one of the driving applications
in the domain of sensor networks. The lifetime of such
systems is envisioned to exceed several years. To achieve
this longevity in unattended operation it is crucial to min-
imize energy consumption of the battery-powered sensor
nodes. This paper proposes Dozer, a data gathering pro-
tocol meeting the requirements of periodic data collection
and ultra-low power consumption. The protocol comprises
MAC-layer, topology control, and routing all coordinated
to reduce energy wastage of the communication subsystem.
Using a tree-based network structure, packets are reliably
routed towards the data sink. Parents thereby schedule
precise rendezvous times for all communication with their
children. In a deployed network consisting of 40 TinyOS-
enabled sensor nodes, Dozer achieves radio duty cycles in
the magnitude of 0.2%.
Categories and Subject Descriptors
C.2.1 [Computer-Communication Networks]: Network
Architecture and Design; C.2.2 [Computer-Communication
Networks]: Network Protocols
General Terms
Algortithms, Measurement
Keywords
Sensor network, data gathering, energy efficiency
1. INTRODUCTION
Observation and interpretation of natural phenomena has
always been of fundamental importance to numerous re-
search areas. Sensor networks represent a tool which pro-
vides the possibility to sample and gather data at scales
and resolutions which were difficult to obtain before. By
spreading large numbers of cheap untethered sensor nodes
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for pro t or commercial advantage and that copies
bear this notice and the full citation on the rst page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior speci c
permission and/or a fee.
IPSN’07, April 25-27, 2007, Cambridge, Massachusetts, USA.
Copyright 2007 ACM 978-1-59593-638-7/07/0004 ...$5.00.
in an area of interest scientists are enabled to monitor dense
temporal and spatial data over an extended period of time.
With this data the analysis of complex interactions becomes
possible; this task is also known as environmental monitor-
ing. Thus, sensor networks have the potential to support
the advancement of various fields of research.
Wireless sensing devices exhibit a large variety of favor-
able attributes. They facilitate the deployment and are far
less intrusive than tethered solutions. Furthermore, they
permit temporary measurements or surveillance of secluded
areas. In addition sensor networks should not need any hu-
man interaction while fulfilling their intended tasks. Due to
the limited capacity of common power supplies for sensor
networks, such as batteries or solar cells, energy efficiency
is a fundamental requisite for prolonged network lifetime.
All sensor nodes are equipped with a short-ranged radio al-
lowing them to convey their data to an information sink for
further processing. This communication subsystem is one of
the primary power consumers of a sensor node. The energy
wastage of the radio, even in idle listening, is three orders of
magnitude higher than a node’s power drain in sleep mode.
As a consequence, the radio should only be turned on if a
data transfer is pending. This requirement is hard to fulfill
since multi-hop routing techniques must be applied to trans-
mit data from all nodes in a possibly large area to the data
sink. Energy-efficient data exchange is a nontrivial task in
single-hop networks but becomes even more challenging if
routing over multiple hops is required. Sensor nodes are no
longer able to schedule their transmissions strictly according
to their individual demands but they also have to activate
their radio in order to receive and relay messages from other
nodes in the network. This raises the well-known problems
of idle listening and overhearing which waste precious en-
ergy.
In this paper we consider applications in the field of envi-
ronmental monitoring—also known as data gathering—for
wireless sensor networks. We thereby focus on applications
producing continuous data. Examples thereof include preci-
sion agriculture [1], glacier displacement measurements [12],
natural habitat monitoring [11], or microclimatic observa-
tions [17]. All of these applications generate periodic data
samples at low rates resulting in light traffic load and thus
low bandwidth requirements. We propose Dozer, an ultra-
low power network stack tailored for data gathering appli-
cations. It incorporates a MAC layer, topology control, and
a routing protocol. We refrained from integrating existing
low-power solutions for any of these subsystems since it is
our strong belief that only a perfectly orchestrated network