Eric Brasseur Home    |    Links    |    Contact    |   

A way to know who listens to what

A key of the spy task is to listen to conversations. Numerous techniques exist to achieve this. Yet in some cases it is difficult or even impossible. This text describes a way to know in such cases what people are listening to, provided that sound is being recorded somewhere else.

Imagine a secret agent is located far from a person he must survey. The person uses a phone. The secret agent cannot hear the conversation. But he knows another secret agent is surveying the correspondent of that person. No matter what they are talking about, a valuable data would be just to know if those two persons indeed did talk together. The two agents must use each a little tape recorder and record while the persons are talking. That even if nothing will ever be audible on the tapes. Then they send the two records to a secret service laboratory. A simple computer program will make calculations on the records... and tell whether the two persons did talk together.

In order to understand how this is possible let's consider a simpler case. You are an agent in front of a building. You are looking to the window of the person you survey. Curtains avoid you see anything. There is few light inside the person's room and it flickers quickly. The person is watching TV. What channel ? This is important data. But you cannot look at his TV. Because of the curtains. Or because of the viewing angle. Yet you have a little pocket TV. Switch it on and tune every channel. Don't look at the little screen. Just let the light it emits fall on the back of your hand, while you are watching the window of the person you survey. Once you tune the correct channel, the back of your hand will flicker exactly the same way as the window. It will be dark when the window is dark, bright when it is bright, flicker quickly when it flickers quickly. You know what channel the person is watching. Even though you don't know what it is about.

Now let's suppose the curtains behind the window are quite thick. You just see a faint flickering from time to time. Then you will have to wait longer till you make sure the channel you tune on your pocket TV is the same. Suppose the curtains are even thicker. Your eyes no longer can notice anything. But there are still very weak flickerings. You can take a camcorder out of another pocket and direct it towards the window. You must record several minutes. Go back to the office and submit the cassette to the laboratory people. They have a record of every TV channel the past hours. A computer will compare the global brightness of the window you recorded with the brightness of the channels at that time. This is simply done by mathematically multiplying the brightness of the window by the brightness of a channel, for every second (more precisely : brightness minus average brightness for both). Then make a sum of those multiplications. The resulting sum will be very high for the channel the person was watching... If the people of the laboratory are not sure of the exact time you made the record, they just have to try out every possible time. The correct time will yield the maximum sum too. This works even if the records of the channels brightnesses were faint too, though a very long record will then be needed in order to get a clear result.

This can be extended to a whole building. You can direct a camera towards the front surface of the whole building, recording the brightness of every window. Then, once the record is submitted to the computer, it will indicate for every window what TV channel the people inside that room are watching. Even if they watch two or more channels on separate TV's. The computer will then show both channels for that window and even tell which channel is being watched on the biggest TV.

This can be done with sound records too. A sound is a signal just like the record of the brightness. Think of the membrane of a loudspeaker that moves back and forth, like the brightness of a screen brightens and darkens. The secret agent can record the sound coming out of the window, instead of the brightness of the windows. By using a directive microphone. And make the computer compare that sound record with the soundtrack of TV and radio channels.

A sound is a signal that changes very quickly, up to 20 000 times per second for human hearable sounds. There can be echoes and some frequencies can be destorted. So in some cases it is better to compare the intensity of different frequencies rather than the sound signal itself. But that's a detail.

There is another difference between a sound signal and a light signal. Sound signals travel through the air at 300 meters per second. So if you are located 300 meters far from the person you record, there will be a delay of 1 second. This will be noticed by the computer that compares the sounds. It will report that in order to get the best match it had to shift the recorded sound forth 1 second. Reciprocally, this allows to know at what distance the person is situated. Imagine you know a person you search for is situated in a neighbourhood, but you don't know in which house. A spy satellite allows you to hear a conversation that person is having with another person. Just record the sound of the neighbourhood, even if strictly no conversation is hearable. And make a computer calculate at what distance the person is situated... After a triangulation, the seek of the maximum signal or the use of the echos, you will be able to know where the person you seek is situated.

Eric Brasseur  -  13 Octobre 2001