Civil Engineering Reference
In-Depth Information
aggregation techniques for other SHM algorithms can be found in Sim and
Spencer (2009) and Sim
et al
. (2010).
It can be seen from the discussion above that model-based data aggregation is
used to distribute the initial part of some SHM algorithms. However, not all the
SHM algorithms can use this technique to reduce wireless communication. For
some SHM algorithms which involve complex matrix computation such as eigen
decomposition, singular value decomposition, model-based data aggregation is not
applicable.
Networked computing seems to be a promising approach to solve the problem.
A wireless sensor network can be viewed as a parallel computer with a number of
processing nodes. The objective then becomes how to perform arbitrary (and likely
complex) computational tasks using a distributed network of wireless sensors, each
with limited resources both in energy and in processing capability. To answer this
problem, it is firstly necessary to find an approach to decompose the complex
computation tasks into smaller operations, each with its own input and output and
collectively related through a certain dependency structure. Then, these smaller
operations need to be distributed among individual smart sensor nodes so as to
incur minimal energy consumption and delay. The research associated with the
networked computing in WSN-based SHM has just started recently. Associated
works can be found elsewhere (Zimmerman and Lynch, 2008, 2009; Jindal, A. and
M. Liu, 2010).
11.4 How to realize fast and reliable delivery of a large
amount of data
SHM generates large amount of data which need to be delivered reliably and,
if possible, within a short period. However, this task is difficult for the following
reasons:
(1) Unlicensed frequency bands have limited bandwidth.
(2) Different layers of network protocols (OS, MAC, routing, network) further
limit the actual bandwidth achieved.
(3) Network size, environmental conditions also affect the bandwidth achieved.
Almost all types of smart sensor nodes use unlicensed frequency bands to
transmit the data. The theoretical bandwidth of an unlicensed frequency band
is generally limited. For example, widely used motes in WSN-based SHM, such as
MicaZ, Imote2, use a 2.4 GHz frequency band. The theoretical bandwidth of this
frequency band is about 250 kbps. Considering the high sampling frequency, this
bandwidth is not adequate for many SHM applications. For instance, for a WSN-
based SHM system including 40 sensor nodes, each generating 16-bit vibration data
at 400 samples per second, the amount of data generated every second is 256 k bits.
This already exceeds the theoretical upper bound of 250 kbps.
Besides the limited bandwidth, network protocols at different layers further limit
the actual bandwidth achieved. In TinyOS, a header will be added in each wireless
Search WWH ::
Custom Search