Compromised Keyboards
It’s enough to send computer-security specialists into a swoon: Could the next threat to corporate secrets be lurking on employees’ own keyboards?
Penn engineering graduate student Gaurav Shah and his colleagues think that’s a distinct possibility, and to prove it, they’ve constructed a device they call a JitterBug, which can be installed on a keyboard to steal and send data over existing network connections. The parts for it cost less than $10.
Shah presented the findings at the USENIX Security Conference in Vancouver, where it was named the Best Student Paper. He credits Dr. Matthew Blaze—his adviser and associate professor of computer and information science—for the idea and the name: “There was this famous bookmaking case in New York a few years ago where … the FBI wanted to access some of [organized-crime boss] Nicky Scarfo’s data, but it was encrypted,” Shah says. “So they had to get permission to break into his house and install this device (called a keystroke logger) that would capture his keystrokes and his password. The problem with keystroke loggers is that you first have to physically install the device, and then you have to go back and retrieve it to get access.”
The JitterBug only needs to be installed; it sends the information over a network by adding imperceptible timing delays, or jitters, whenever the user presses the computer keys. JitterBugs could be installed on a computer mouse, a Web cam, or any other input device—even during an earlier stage of the assembly process, Shah says. The prototype he created using “off-the-shelf” parts is a bit smaller than a cell phone, but it could be made even smaller and less noticeable.
Before people start signing up for JitterBug lessons, they should note that Shah views JitterBugs as a threat for companies, agencies, or universities trying to prevent leaks of sensitive data, not individuals. “Home users definitely don’t need to worry about this.”
Measuring Moods from 9 to 5
First there’s the fight over breakfast about the credit-card bill. Then comes the speeding ticket during the morning commute. Will this workday be a wipeout?
That’s the kind of question Wharton management professor Nancy Rothbard was interested in answering when she and Ohio State University researcher Steffanie Wilk examined the effects of start-of-day mood on employees at an insurance company’s call centers. “One of the things I always wondered about was how emotions affected performance and how what you brought with you might affect your emotions throughout the day,” Rothbard says.
It turns out that start-of-day mood is powerful and persistent: “Your starting point matters.”
Several times a day, for three weeks, customer- service representatives answered computerized questionnaires to assess their own moods as well as the moods of their customers. Rothbard and Wilk also looked at daily work-performance measures used by the employer.
They found that people who started work in positive or negative moods generally stayed that way throughout the day. The study also found a slight effect of mood on work performance: Employees reporting happy starting moods took fewer breaks and transferred fewer calls to their supervisors. Unhappy workers handled fewer calls and reported being less engaged in their work.
Rothbard speculates that starting mood may provide a lens through which one views the rest of the day’s events and interactions. Another possible explanation is that “what you bring with you to the workplace is an essential part of your identity and your life. If you have a fight with your spouse, you may be thinking about it all day.”
Rothbard was surprised to find that negative customers had little effect on the call-center employees’ moods but that cheerful customers tended to boost their moods. People who work in these jobs are trained to deal with difficult customers and may be immune to the effects of an unpleasant phone exchange, Rothbard speculates. In addition, “customers come and go and they’re not so much a permanent part of the constellation of difficulties that people carry around in their lives.”
The Eye and the Ethernet
When Penn scientists put a guinea-pig retina under the microscope and showed it movies of natural scenes representing four types of biological motion, its responses allowed them to estimate the rate at which visual information is transmitted by the human retina. At about 10 million bits per second, “You could say it’s on the order of something like an Ethernet connection,” says Dr. Peter Sterling, professor of neuroscience and senior author of a study published in the July issue of Current Biology.
The study was a collaboration between Penn’s neuroscience department, including lead author and Ph.D. student Kristin Koch; Princeton University’s molecular-biology department; and Dr. Vijay Balasubramanian, Penn’s Merriam Term Associate Professor of Physics, who led the analysis.
The retina is a part of the brain that has grown out into the eye, and Sterling says that gram for gram, it’s 50 percent more “expensive,” in terms of its energy use, than the average part of the brain. The guinea-pig retina has about 100,000 ganglion cells that process different types of visual stimuli and work together to send a complete picture to the brain through what are basically digital signals. (The human retina, in contrast, has about one million of these cells.)
For this experiment they placed the flattened guinea-pig retina on top of a multi-electrode array that recorded the electrical spikes fired per second in dozens of cells at the same time. The researchers then projected such images as a swimming salamander down the camera port of the microscope.
The Penn scientists studied seven types of ganglion cells and classified them as “brisk” or “sluggish,” according to the rate of electrical spikes fired in response to the stimulus.
Until now, almost all of the research has been done on one or two types of cells that are large and fire rapidly, Balasubramanian says. “But actually 90 percent of the work is carried out by these other cells that nobody studies.”
While Sterling’s main goal is to continue shedding light on the brain’s circuitry, the data could also be useful in the design of artificial retinas.
“Whatever you do to make an artificial device,” he says, “we now know about how much information you need to transmit to make it comparable to a biological device.” —S.F.