scattered light wave. This shift is affected
by longitudinal distortion along a cable,
which, in turn, is affected by strains in the
fiber arising from interaction with the surrounding soil. The system delivers continuous information from the fiber, and a single device can monitor up to 30 km of
fiber optic line. That makes for low-cost
detection because the cost of standard
single-mode communication fiber runs a
fraction of a dollar per meter and one-third
that per foot.
“…the detection rate
is high, regardless
of tunnel size.”
However, the data from the analyzer, by
itself, does not distinguish aboveground
activities such as rain or soil loading from
underground ones; e.g., tunneling. To
make the distinction, the researchers decomposed the complex measured signals
into simpler wavelets. They used analytical solutions and numerical simulations to
train an automatic detection system based
on a neural network, training and validating it with about 50,000 simulations. The
resulting system requires no special
expertise to operate and does not need a
constant presence of personnel along the
perimeter, unlike alternatives such as
ground-penetrating radar.
The scientists adjusted the neural net to
avoid false alarms, a key criterion if the
system is to be implemented in a dangerous situation. The trade-off is a greater
likelihood of not detecting a tunnel.
In practice, the fiber fence could be
arranged in two configurations: buried at a
shallow depth of a meter or so, perhaps
running parallel to an aboveground fence,
or buried in shafts, separated by less than
20 m.
According to the researchers, a prototype tunneling detection system could be
deployed within a few months. Moreover,
the technique could be used for other applications after refinement and further development. The researchers are investigating some of these alternative uses.
“We have been working on a similar
system to detect sinkhole development,
which is a major problem in the Dead Sea
area,” Klar said.
Hank Hogan
hank@hankhogan.com
