Tuesday, April 21, 2015

Meeting #5 - Recap

"Locating Intrusions Using Fiber Optics"

For its fifth meeting, EAST Knowledge was pleased to have Jim Rulla as its guest speaker. Jim previously spoke at EAST Knowledge for our second meeting, and he was gracious enough to come back. The topic of this meeting was “Locating Intrusions Using Fiber Optics”. As usual, we’ll do our best to report on the talk, but these summaries are in no way intended as a substitute to seeing the talks in person.

Jim, always an engaging speaker, opened his talk by showing the audience a front page news story from 2012 about a man whose jet ski had broken down while riding it New York's Jamaica Bay. The man swam three miles to shore and climbed the fence at the JFK airport without detection. It was a dramatic example of the importance of perimeter security, and an example of where the technology Jim was about to talk about could be applied.

Basics of Interferometry

Jim gave a basic description of interferometry. Interferometry is a way to perform high precision measurements on the order of wavelengths of light. Interferometry relies on the wave properties of light. When multiple waves are combined, they add together. This adding together can be constructive (when both waves are positive), or destructive (when one wave is positive and the other negative). A basic interferometer uses a laser source, a beam splitter, two mirrors, and a detector. Example from wikipedia:

The key to the interferometer shown above is that the light travels along two different paths but are recombined in the detector. Now if you move one or both mirrors, the path length of the two light beams will change. This will cause fringes to appear at the detector as the two light beams will be slightly out of phase with one another. If you are varying the position of the mirrors in time, the fringe patterns will also vary in time. Example fringe patterns:

How Interferometry can be Used to Detect Intrusion

Jim had his custom software that he used to demonstrate the basic sort of signals that would be generated by a an interferometer. In the picture you see below, there are two large plots. The large plot on the bottom represents the position of one of the mirrors over time. The large plot on the top represents the signal that is detected where the two light beams come back together. As the mirror moves up and down (changing the path length the light travels), you can see the affect on the brightness of the light. Notice that when the mirror isn’t moving, the signal is flat (left side of plots).

Top plot represents interferometer signal. Bottom plot is the mirror position. Notice that when the mirror isn’t moving, the signal is flat (left side of plots).

To detect an intrusion using interferometry, you would need an interferometer where an intrusion would disturb the path length of light. You can create an interferometer using fiber optics. If you then run the fiber optics along a fence (for example), any disturbance on the fence will cause a disturbance in the fiber optics. This in turn changes the path length of the light, which will create interference patterns that are then detected. In the absence of an intrusion, you have a flat signal. Once an intrusion occurs, the signal begins to change rapidly. Detecting when the signal is changing is all that is needed to detect the intrusion.

Jim also mentioned that basic interferometry makes it difficult to determine which direction the path length has changed. You’ll notice in the plot shown above that as the mirror changes direction, the output of the basic interferometer continues to change in the same way. However, modulating the light being used from the interferometer provides phase information. One possible way of doing that is by vibrating one of the mirrors. This technique is known as "homodyne modulation".

How Interferometry can be Used to Detect Location

Now that we have a basic interferometer detection, how can that be used to detect location? Jim showed a diagram (poorly) captured in the photo below. In the top, you see two fiber optic interferometers (one black the other red). These interferometers run in opposite directions in the same bundle of fiber. Because the interferometers are in the same bundle, what happens to one interferometer also happens to the other. If an intruder is climbing a fence, he will disturb both interferometers.

Because the interferometers are running in opposite directions, the signal produced by one interferometer is displaced in time relative to the other interferometer (see the red and black step plot in the picture below). This time difference varies depending on where the intrusion occurs. If you can calculate the time difference, then you can determine the location of the intrusion because you know the speed of light. Pretty cool!

In order to determine the time difference you have to perform a correlation on the two signals. The correlation isn’t perfect and can be computationally expensive, but using the correlation, you can get an estimate of the time difference between the intrusion signals.

The two interferometers running in opposite direction are shown on top. In the middle you see a simple plot showing the interferometer signals from a single intrusion. Notice that the intrusion signals are offset in time.

Jim showed that with this method of determining location that there is a direct relationship between the speed of light and how rapidly you would have to sample the signals. For example, to try to get an accuracy of +/- 25 m, you have to sample the interferometer at least 10 MHz. He further emphasized that if you have a typical DSP running at 500 MHz, that you would only get 50 clock cycles per sample. That means that to properly process the incoming data, you need to have the appropriate model and the appropriate, efficient approach. Using his software, Jim showed off several different approaches to processing the signals. In one example he emphasized the processing speed. In another he emphasized the accuracy and certainty of the location calculation. He suggest that for this particular application it is possible to get both.


Both during and after the talk, the audience had a lot of great questions and made several astute observations. We also have some leads on possible future talks. For anybody who might be interested in discussing some tech topics that interest them, get in touch. You’ll find a receptive audience with EAST Knowledge. If you’re eager to share, we’re eager to learn.

We hope to see you next time for stimulating conversation. (We also have coffee, cookies, and Raspberry Pi.)

References and Resources:
  1. Link to jet ski story: http://abcnews.go.com/US/jet-skier-broke-jfk-airports-security-wanted-caught/story?id=17550202
  2. More on interferometry: https://en.wikipedia.org/wiki/Interferometry
  3. More on homodyne modulation: http://en.wikipedia.org/wiki/Homodyne_detection
  4. More on correlation: https://en.wikipedia.org/wiki/Cross-correlation
  5. Patent on Intrusion Location: Apparatus and method for monitoring a structure using a counter-propagating signal method for locating events
  6. Jim's talk for meeting #2, announcement and recap:
    1. Announcement: http://www.eastknow.org/p/events.html#meeting-2
    2. Recap: http://www.eastknow.org/2014/03/meeting-2-recap.html
  7. Wikipedia images:
    1. Interferometer: https://commons.wikimedia.org/wiki/File:Interferometer.svg
    2. Fringes: https://commons.wikimedia.org/wiki/File:Colored_and_monochrome_fringes.png

Saturday, April 4, 2015

Meeting #5 Announcement - Locating Intrusions Using Fiber Optics

Update: A meeting recap.

EAST Knowledge is excited to announce it's fifth meeting. Jim Rulla will be giving a talk about how fiber optics can be used as sensors for locating intrusions. Come and learn something new! Come and share what you know! All are welcome!

What:Locating Intrusions Using Fiber Optics
Who:Jim Rulla
Where:The Story Room at the Gresham Library (map)
When:10:30am-Noon, Saturday, April 18th, 2015
Why:Sharing what we know

Interferometers measure distances to within fractions of the wavelength of light. Put a couple of these incredibly sensitive devices on the fence around an airport, and you can not merely detect a would-be intruder — you can locate him. Put a pair on a pipeline, and you can not merely detect nearby digging that could damage the pipe — you can locate its source. You might even be able to detect — and perhaps even locate — leaks in the pipe.

The detection and location mechanisms are fascinating examples of clever engineering. I'll explain how these patented systems work and illustrate new algorithms that improve performance.