Yao-Win Hong, Birsen Sirkeci-Mergen, Anna Scaglione, Rajit Manohar
In this paper we compare the energy efficiency of wireless
sensor networks sampling a continuous sensor field in two
different types of deployment, which we refer to as the High
Density (HDSN) and the Low Density (LDSN) Sensor Network
architectures. In the LDSN, a set of sensors with high
resolution are critically deployed at sampling locations so
that each sample is nearly uncorrelated and is transmitted
to the central node in a separate channel. In HDSN, a simple
zero-crossing detector is used at each sensor and the
sensor field is reconstructed at the central node with the
zero-crossing information extracted from the sensors' observations.
By proposing a scalable data collection protocol
for HDSN, we show that the reconstruction performance
of the sensor field at the central processor can be achieved
with low complexity at the same bandwidth and energy cost.
Therefore, the longevity of the sensors is increased due to
the reduced per node energy consumption and the reduced
computational energy for the data representation at each
sensor. Furthermore, we claim that the system versatility
and fault tolerance of HDSN makes it an better alternative
to the LDSN architecture.