Loose clicks sink computers

Phenomenon: Stray signals emitted by electronics can reveal secrets.

Medicine & Science

July 19, 2004|By Michael Stroh | Michael Stroh,SUN STAFF

VISITORS TO Markus Kuhn's laboratory arrive eager to be duped. And the Cambridge University computer scientist is happy to oblige.

Ushering a guest into an neighboring room, Kuhn asks him to pull out his laptop computer, wait until he's gone, and then tap out a sentence or two. A few minutes later, Kuhn returns and tells his visitor exactly what he wrote. "It's usually quite impressive," he says.

Kuhn is one of a handful of researchers probing a James Bondian borderland of computer science: a phenomenon known as "compromising emanations." The term refers to stray signals emitted by a computer or other electronic device that can inadvertently betray sensitive information.

Long of concern to the U.S. military, which operates a classified program to squash them, compromising emanations are a growing draw for civilian computer scientists.

"I think emanations are extremely overlooked," says Avi Rubin, technical director of the Johns Hopkins University's Information Security Institute. They're not only technically fascinating, he says, but an important, if exotic, potential threat.

Although most people never notice them, computers can leak a variety of indiscreet signals. In May, IBM researchers announced they could figure out what someone was typing from the clicks made by the keyboard. The human ear can't tell the clicks apart. But researchers discovered that a cheap microphone hooked up to a computer can.

Kuhn's digital magic act, on the other hand, takes advantage of radio waves leaking from laptop video connectors. (Cathode ray tubes inside bulky computer monitors do the same.) As the military has known for decades, these signals can be snatched and decoded.

Kuhn says emanations he and others have ferreted out are just the beginning. "There are probably a half-dozen or dozen exciting phenomena yet to be discovered," he says.

While they sound like a modern dilemma, compromising emanations have a long history in the annals of military espionage. In his doctoral thesis published in December -- widely considered the most comprehensive unclassified treatise on emanations -- Kuhn describes how governments grew interested.

It happened during World War I, when the German army discovered that the primitive battlefield telephone lines used by British and French troops emitted electromagnetic signals that could be detected from a distance. By the time the British and French figured out the technique, the Germans had learned to shield their phones.

Compromising emanations proved useful during peacetime as well. In the book Spycatcher, former British security service scientist Peter Wright recounts how British agents eavesdropped on French diplomats during tense negotiations over Britain's entry into the European Economic Community. Britain's source: stray radio waves emanating from the French embassy's telex cable.

In the 1960s, the U.S. military launched a top secret effort, code-named Tempest, to develop equipment that could detect and block compromising emanations. The program, still run by the National Security Agency, led to countermeasures such as wrapping sensitive computers, rooms, even entire buildings, in copper or stainless steel foil.

In 1985, Dutch computer researcher Wim van Eck published one of the first nonclassified research articles on compromising emanations in the technical journal Computers & Security. Later that year, van Eck orchestrated a demonstration of the phenomenon for the BBC program Tomorrow World.

Sitting in an antenna-equipped van outside a London office building, van Eck was captured on television as he tuned into radio waves emitted by an unshielded computer operating inside. What made the program riveting to British viewers was not just the technology, but where van Eck was parked: in front of New Scotland Yard.

Although electromagnetic radio waves have traditionally been the biggest security concern, in recent years creative researchers have found new, less obvious electronic leaks.

Using less than $200 worth of equipment, Dmitri Asonov and Rakesh Agrawal, computer scientists with IBM's Almaden Research Center in California, figured out how to translate clicks made by a computer keyboard into the letters on keys that made them.

The clicks occur when keys strike a plastic membrane sandwiched between the keyboard and its base. The membrane, researchers found, behaves like the skin covering a drum: Depending on where the key strikes the membrane, a unique sound wave is produced.

While its indistinguishable to the human ear, Asonov and colleagues were able to train neural network software to read key clicks with nearly 80 percent accuracy. By switching from a cheap microphone to a parabolic mike, the researchers were able to eavesdrop just as accurately from nearly 50 feet away. Their neural network software also worked on telephone and automated teller machine keypads.

Baltimore Sun Articles
Please note the green-lined linked article text has been applied commercially without any involvement from our newsroom editors, reporters or any other editorial staff.