Christian ENZ
Title of the Talk: Low Power Wireless Vision Sensors Network for Context Awarness
Abstract: One of
the important challenges faced when building wireless sensor networks
(WSN) is
to reduce the power consumption in order to achieve a sufficiently long
node
lifetime (typically 5 to 7 years on a single AA alkaline battery).
Although the
MAC layer plays a crucial role in the overall energy efficiency, the
radio
remains one of the bottle-neck for implementing ultra low-power WSN.
The power
consumption of the radios available today does not allow for continuous
operation and the radio has to be duty-cycled in order to reach the
targeted
several years autonomy. This clearly has an impact on how to design a
radio for
WSN. As an example, the WiseNET ultra low-power radio developed at CSEM
will be
presented. It combines a dedicated duty-cycled RF CMOS radio with
WiseMAC, a
low-power MAC protocol designed for WSN. The WiseNET solution typically
consumes more than 30 times less power than comparable solutions
available
today, using for example IEEE 802.15.4.
Visual
scenes can provide a lot of information that can be used to enhance the
user
awareness to his surrounding environment. This can be done with a
wireless
network of distributed cameras, where each camera transmits its video
stream to
a remote computer where the video signals are processed and the global
visual
information is extracted. This approach requires a high bandwidth and
is
therefore not well suited for a low-power wireless implementation.
Furthermore
it requires a huge amount of computation that makes real-time operation
almost
impossible to achieve. A better approach is to use vision sensors in
place of
traditional cameras. Vision sensors are based on the fact that,
statistically,
most of the relevant information of an image frame is contained in the
largest
contrast amplitudes, which represent only a small fraction of all
pixels. Using
this contrast information, the features of interest can be extracted
locally
and communicated to the user or shared with other vision sensors over a
low
data rate wireless link. This example clearly illustrates the trade-off
between
the energy expensive RF wireless communication and the local digital
signal
processing, which becomes more and more energy efficient thanks to the
down-scaling of CMOS technology.
This
talk
presents the recent results achieved within the WiseNET project running
at CSEM
in the field of wireless vision sensors network applied to context
awareness.
It describes how the ultra low-power WiseNET radio could be used
together with
the vision sensor platform to build wireless networks that can be used
for
extracting information from the user environment. Several examples in
the field
of building control (people tracing within buildings, intrusion
detection,
lighting control,...) and car driving assistance (lane departure
warning, seat
occupancy detection,...) will be given.
|
