Two US researchers believe they have found a way to transmit information safely over an optical network without fear of interception. The technique hinges on transmission of encrypted data in the "noise" of signals along fibre-optic cables.
Their method take advantage of the fact fibre-optic systems inevitably have low levels of "noise" — the random jitters in the light waves that are used to transmit information through a network. The technique, developed by two researchers at Princeton University, hides the secret encrypted message in this optical noise.
The sender first converts the secret message into a short pulse of light. Commercially available Code Division Multiple Access (CDMA) equipment spreads the intense, short pulse into a long, faint stream of optical data. The breakthrough comes from the ability to make the signal fainter than the noisy jitters in the fibre-optic cable.
The intended recipient decodes the message by using information about how the secret message was originally spread out, using another optical device to change the message back to its original state. The method should be secure because even if eavesdroppers knew a secret transmission was taking place, any slight imperfection in their knowledge of how the secret signal was spread out would make it too hard to pick out amid the more intense public signal.
The technique was developed by researchers Wu and Evgenii Narimanov of Princeton University and was presented this week at the annual meeting of the Optical Society of America: traditionally a showcase for new ideas. The paper was published originally on 1 May.
Talkback
it's going to be very interesting to see how well this plays out...
Old news and not accurate ... This type of trick people have been trying since World War II with normal radio waves. Since you can easily detect, with the right equipment, the higher energy densities of parts of the spectrum that contain the extra information it will not work. Only if you use ridiculoulsy low data data densities in the stream will you be able to truly stay unnoticed from a statistical c.q. mathematical point of view. The data throughput bandwidth will however become almost unusable.
Anyone with a decent background in signal analysis can see this. It is based on an old idea regarding simple spread spectrum techniques with data "hidden" in the noise. Using Fourier Analysis you can however detect the higher energy levels. Detecting which peaks are part of the datastream is another thing offcourse. So Stealth it is not.
Is it encryption?
No. It is security through obscurity. The security is based on the fact that you would need the key to know at which frequencies to listen at what time and for how long.
Can this be cracked?
If you have the money to listen to all frequencies simultaneously using a sampling rate that is faster than the minimum length of a signal that transmits 1 bit. So yes in theory you can crack such a system.