Let chaos keep your secrets safe


To recreate the chaos you have to send the chaotic light signal made by your first laser, complete with encoded message, to a second laser that is essentially identical: made from the same batch of semiconductor by the same manufacturer and operated at the same temperature and current biases and with the same fraction of light fed back into the cavity (see Diagram).

Enlarge image

Let chaos keep your secrets safe
• 19 November 2005
• From New Scientist Print Edition. Subscribe and get 4 free issues.
• Michael Brooks
More Info-Tech Stories
More Stories
THEY are known as the Squidgy tapes. These covertly recorded phone calls between Diana, the Princess of Wales, and her friend James Gilbey provided the world's media with page upon page of salacious speculation about their relationship.
But you don't have to be in the public eye to be concerned that your calls might be intercepted. A government report released this month revealed that the number of mistakes made by the UK's security services when applying to tap people's phones rose more than sixfold in the last year, with most of the mistakes involving an attempt to tap the wrong number. With fears of terrorism loosening restrictions on what the authorities are allowed to do, even the most law-abiding citizen could be forgiven for wondering who is listening in to their private phone conversations.
Against that background, a study published in Nature this week could spell good news for everyone except spies and gossip-seeking journalists. It shows that keeping calls confidential needn't involve arcane and complex technologies such as quantum cryptography. Existing communications networks will work just as well. The key is adding in chaos - specifically, the strange phenomenon of the butterfly effect.
When Edward Lorenz described this aspect of chaos theory in 1972, he probably had no idea what he was starting. In a talk entitled "Predictability: Does the flap of a butterfly's wings in Brazil set off a tornado in Texas?" Lorenz described how the long-term behaviour of chaotic systems such as the weather are so dependent on minute, usually inaccessible, details in their make-up that they are essentially unpredictable.
But crucially, they are not random: they actually do display some patterns. Within any particular system, the physical variables (such as wind speed, air pressure, humidity, temperature and ocean current speeds for weather) are interdependent. Increase ocean temperature, for instance, and you affect the humidity and temperature of the air above. Steve Strogatz, a mathematician at Cornell University in Ithaca, New York, likens the variables to improvising dancers who continuously react to the unanticipated moves of their partners. While each dancer is free to move any way they want, they will watch each other and coordinate their movements.
Lou Pecora of the US Naval Research Laboratory in Washington DC was the first to recognise that this hint of order within chaotic systems might allow chaos theory to be put to work. In the mid-1980s, Pecora was fascinated by chaos, but his bosses needed to hear more than "it's interesting" before they would let him loose to work on it. Chaos described the unpredictability of systems in the natural world; what good was that to the US navy? But it didn't take Pecora long to realise that this unpredictability might be useful for hiding messages. Bury a message in the chaotic signal from a suitable electrical circuit, and no one would be able to access the message unless they could see through the chaos.
The idea is essentially a chaotic version of a radio transmission. AM radio signals, for example, encode music or speech on top of a "carrier" wave of a particular frequency. Listeners tune their radio receivers to that frequency, and the electronics in the receiver subtracts the carrier. The chaotic radio signal, Pecora reasoned, could work in exactly the same way, with the chaos acting as the carrier wave.
Except that you can't tune in to chaos as you can to an ordinary carrier wave. If you want to send a message buried in a chaotic signal, the receiver has to know the exact pattern of chaos used to encode the message. "You need a very well-matched transmitter and receiver," says Alan Shore of the University of Wales at Bangor, part of the international team of researchers behind the recent work, which was carried out in Athens, Greece. "Extracting the information requires chaotic synchronisation."
The idea of synchronised chaos seems counter-intuitive. What makes it possible is the patterned way in which a chaotic system dances around its possible states. Just such a system can be made from a chunk of light-emitting semiconductor, the material that lies at the heart of every solid-state laser. Laser beams are usually thought of as the most ordered form of light, consisting of streams of rigidly coordinated photons of the same phase and frequency. But if a small fraction of a solid-state laser's output is split off and fed back into the laser's cavity, it stimulates the atoms in the semiconductor to produce more and more light. Pretty quickly the material begins to behave like a loudspeaker in overdrive, producing light's equivalent of feedback noise - a chaotic mix of different frequencies with wildly varying amplitudes.
The light this system produces looks like noise, Shore says. And to people trying to hide their messages, noise is just what they would like their signal to look like - so long as that "noise" can be reproduced exactly by the receiver.
Wanted: laser twins
But that isn't straightforward. To recreate the chaos you have to send the chaotic light signal made by your first laser, complete with encoded message, to a second laser that is essentially identical: made from the same batch of semiconductor by the same manufacturer and operated at the same temperature and current biases and with the same fraction of light fed back into the cavity (see Diagram). Get this right, though, and the chaotic input acts like one of Strogatz's rogue dancers, forcing the receiver laser's physical variables - the energy states of the semiconductor - into performing the same "dance" that the transmitter's semiconductor is performing. In other words, their chaotic light beams become synchronised.
And that is exactly what is required for chaotic communication. Add a message onto the original chaotic light, and the receiver will be able to reproduce the chaos of the carrier and subtract it from the signal it receives to reveal the message.
The first people to demonstrate that this would work were Gregory VanWiggeren and Rajarshi Roy at the Georgia Institute of Technology in Atlanta. In 1998, they sent chaotically encoded messages across their lab at the rate of 150 million bits per second (Science, vol 279, p 1198). Several groups have since carried out similar experiments in the lab, but until now no one knew whether it was possible to do the same thing using the standard optical fibres in a city's telephone system. "There have been questions over whether effects within optical fibres would stop it working," Shore says.
The Athens team has provided the answer. Using just such optical fibres, they sent a chaotically encrypted message 120 kilometres at data rates similar to those used by telecoms companies. And that is why the experiment is so significant: it shows that if this kind of encryption system were rolled out tomorrow it would be compatible with the optical-fibre infrastructure that carries thousands of telephone calls down a single fibre. The Athens experiment did not even optimise the technology, Shore says: the system used standard, off-the-shelf components.
Though there are many questions still to be answered, this system could eventually be used by telecoms operators, with minimal changes to their networks, to provide additional privacy over existing communication links. Alternatively, they could offer chaotic encryption on dedicated networks: banks could use it to back up data and businesses could use it on private data networks. It could also prove useful to security services, Shore says. The same idea could provide encryption for cellphone signals and might also be useful for tasks such as sending secure signals to reprogram satellites.
In fact, the researchers have their eyes on many of the applications that quantum cryptographers have talked about. Quantum cryptography uses the quantum states of light to encrypt data, and this is a relatively slow process which puts a brake on the rate at which data can be sent (New Scientist, 2 October 1999, p 28). Chaotic cryptography, by contrast, can send data at around 2 gigabytes per second, which is the sort of speed used in standard optical telecommunications.
However, no one is yet claiming that chaotic encryption will necessarily spell the end for its quantum-based cousin. Quantum cryptography is inherently secure because quantum states are so delicate that any attempt to listen in disrupts the message, alerting both sender and receiver. While chaotic cryptography can provide impressive levels of security, it does not have the same built-in resistance to eavesdroppers.
Typically, the message is injected into the chaotic carrier at about 1 per cent of the chaotic signal's amplitude. But even if the amplitude is higher, Shore and others have shown that an eavesdropper could not pull the message out without exactly the right equipment. First they would need an optical coupler to bleed off some of the chaotic light. Then they would need to attach the right kind of laser - to match the device used to create the message - and find exactly the same operating conditions.
Alternatively, if they wanted to use computers alone to crack the code, they would need to record the transmission with picosecond precision. The high bit rates would help thwart any such attempt, as most eavesdroppers would struggle to cope with the data stream. And efforts to extract the message would be complicated by the fact that a computer would need to digitise and hence distort the analogue signal. Claudio Mirasso from the University of the Balearic Islands in Majorca, Spain, who was part of the Athens team, believes that this task would be almost impossible. The Athens team employed people to try to break into their chaotic encryption, but no one succeeded.
But as in all areas of cryptography, a code that has not yet been broken is no guarantee of security. In the earlier lab tests of chaotic cryptography, code breakers were able to bust the code wide open. As part of a project funded by the US National Security Agency, for example, Kevin Short, a mathematician at the University of New Hampshire, not only showed that they were all breakable in principle (Physical Review Letters, vol 83, p 5389), but also broke every code that was thrown at him. The key, Short says, lies in the scheme's reliance on synchronisation, which in turn depends on patterns in the chaotic signal. And patterns are a weak link in any cryptography scheme.
“The researchers have their eyes on many applications that quantum cryptographers have already talked about”
When the chaotic signal enters the receiver laser, it forces all of the receiver's physical variables to dance in particular ways. But there is part of the signal - the hidden message - that does not dance in the same way because it is not associated with the same physical characteristics. And so it stands out. "You can break in because the messages won't be correlated with the chaos," Short says. "Even though data looked really chaotic, we could extract the signal." Short believes that this problem makes chaotic encryption vulnerable.
Mirasso accepts that chaos-based encryption systems are not 100 per cent secure. But he says Short was given some information that would not be available to an ordinary eavesdropper. And by using more advanced lasers that behave in an extremely non-linear way, he believes that chaotic encryption could prove highly secure. Short concedes, in turn, that the fact that he could break the chaotic codes doesn't mean they are useless. "I think these things are great schemes, and there's a great deal of brilliance in them," he says. "If you need security for a number of hours, where it wouldn't be easily detectable, or where there's a low probability of interception, it's probably useful."
“Routine encryption of phone calls would be welcomed as it could make casual line-tapping a thing of the past”
Strogatz agrees. "For certain purposes you don't need tremendous security; just a little privacy would be welcome." It could be useful to police during a sting operation, for instance. Rather than maintaining radio silence, they could use chaotic encryption to make their communications look like noise: anyone trying to listen in with a scanner would be unaware of any messages travelling through the airwaves.
Devising a means to measure the security of their system is an important next step for the Athens researchers. "The question is, what do you need to break the code?" Shore says. "It's very difficult to quantify that." All he can say at the moment is that the high bit rates make the chaotic code "a significant challenge".
The project might have spin-offs outside encryption and telecommunications - in biology and medicine, for example. Researchers are still debating how neurons move information around in the brain; perhaps information could be encoded in the rates at which neurons fire, or in the timing of firing, Shore points out. The underlying difficulty in finding out lies in the complex nature of biological signal carriers. "That complexity is shared by optical chaos communications," Shore says.
People with epilepsy might also one day benefit from this line of study. Researchers have identified a chaotic complexity in patterns of brain activity immediately before a seizure. This, Shore says, comes from the synchronisation of groups of neurons that are usually independent of one another. Understanding how neurons synchronise during a seizure might give warning, and thus a chance to medicate against the attack. Researchers might even be able to learn how to use some kind of electrical stimulation that would stop the synchronisation in its tracks.
In the end, the power of synchronised chaos will be judged on the applications that use it. The routine encryption of phone calls is certainly something that civil liberties campaigners would welcome, as it could make casual line-tapping a thing of the past. And if it really can help us prevent epileptic seizures, there's no doubt that will be explored, too.
In their original experiments, Shore and his colleagues weren't trying to address acute medical problems, they were just trying to get chaotic communication off the ground. But that's the butterfly effect for you. Start flapping around in one area, and who knows what you will kick off elsewhere.
From issue 2526 of New Scientist magazine, 19 November 2005, page 32

Posted: Mon - November 21, 2005 at 11:13 PM          

©