Maximum Connectivity

By Denise DiFulco

“I’ve always been motivated by research that has an everyday use in real life.”—Jeannie Albrecht

Dropped calls have been a nuisance for about as long as cell phones have been in existence. But with mobile devices handling increasingly complex operations—from text messaging to video chats to emails with huge attachments— network connectivity is more important than ever.

It’s an issue that computer science professor Jeannie Albrecht is working to improve. In 2009 she was awarded a $400,000 National Science Foundation Career Award (rarely given to small, liberal arts colleges like Williams), which she is using to develop computer programs that keep devices continuously connected to mobile networks—or at least trick them into believing that they are—to avoid service disruptions.

“We’re trying to mask any time you’re disconnected from the network,” she says, explaining the software she has been designing with the help of students and independently. “We’re trying to develop something that will run below your applications but above the operating system.”

Albrecht is currently studying how power loss and malfunctioning equipment—two of the many factors that can disrupt a wireless computer network—affect the University of Massachusetts Amherst’s DieselNet, a system of city buses outfitted with computers that are connected by radios. Disruptions can occur when the buses power down at night or while they are moving, accessing different wireless points along their routes.

Albrecht and her students have written software that monitors network activity for these kinds of hiccups in the system and then prevents applications from noticing similar disruptions in the future.

One solution she has investigated is delay-tolerant computing. When a bus becomes disconnected for a brief period of time, for instance, a delay-tolerant computing framework will essentially act as a buffer, preventing the application from experiencing any disruption in service or connectivity.

Another critical part of her research is “partially predictable” resource scheduling. That is, it may be possible to anticipate the routes of the buses and their associated network connectivity to schedule the programs to run in a predictable way.

Albrecht says the immediate payoff of her inquiry will be for fellow researchers. She wants to set up a software layer that sits above an unreliable mobile network to handle failures and allow for more trouble-free testing and development.

Her work eventually could lead to the creation of algorithms and programs that are more broadly applied to the electronic devices that we use every day. “Even laptops move from place to place. Cell phones. iPods,” she says.

As an outgrowth of her research, Albrecht is also looking at partially predictable resource scheduling as a way to reduce peak energy usage in the home. “If we can predict how much energy the coffee pot and the toaster are going to draw in the morning, it may be possible to postpone the dehumidifier from running in the basement at the same time,” she explains. Reallocating energy usage in that way could lighten the load on power grids and save consumers money at peak times.

Before coming to Williams in 2007, Albrecht received a master’s degree in computer science at Duke University and a Ph.D. at the University of California, San Diego. She enjoys a variety of activities that are decidedly “low-tech,” including Ultimate Frisbee, skiing, surfing and kayaking. Her inclination to the outdoors may feed into her research, as she is eyeing other types of test beds as well, including sensors deployed in river ecosystems and networks of wireless sensors used in urban settings to monitor weather and air pollutants.

Thanks to the National Science Foundation grant, Williams students are at the forefront of Albrecht’s testing and software development. A recent group of her Winter Study students created and released the HungryEph app for the iPhone, iPad and iPod to help users maximize dining-hall meal points. And she recently completed work with thesis student Danny Huang ’11 that focused on improving smart phone performance in areas where network connections are spotty—like along Route 2 through the Berkshires.

At a larger institution, a Career Award might partially fund one or two graduate students for the duration. But Albrecht says she will be able to fund at least one student every summer for the next five years. “It’s really quite generous for a small school,” Albrecht says. “It was a nice surprise.”

IMAGE CREDIT: Photography by Jon Roemer