Albrecht shares vision for efficient smart homes

Jeannie Albrecht was the first speaker in this year’s Faculty Lecture series, and she spoke on automating smart homes. (Photo courtesy of Williams College)
Jeannie Albrecht was the first speaker in this year’s Faculty Lecture series, and she spoke on automating smart homes. (Photo courtesy of Williams College)

The College started its 2014 Faculty Lecture Series on Thursday with a talk by Associate Professor of Computer Science Jeannie Albrecht titled “Detecting and Predicting Home Occupancy in a Smart Home.”

The faculty lecture series was founded in 1911 by a member of the College community whose goal was to “relieve the tedium of long New England winters with an opportunity to hear Williams professors talk about issues that really mattered to them.” This year’s lecture series was chosen by a committee of three professors: Professor of German Helga Druxes, Associate Professor of Mathematics Mihai Stoiciu and Assistant Professor of Africana Studies Neil Roberts.

Albrecht’s talk was on how smart homes can be used to reduce energy consumption. In the last 20 years, the energy demand in the United States has increased by 50 percent. But approaching energy issues is extremely difficult because “energy generation and energy consumption are extremely fragmented,” Albrecht said. Buildings consume approximately 48 percent of our nation’s energy, and the remaining half is divided between industry and transport.

The goal of Albrecht’s research is to “reduce consumption by making buildings smarter and greener.” Smart houses will reduce waste, cost and environmental impact. The obvious solution, according to Albrecht, is to use less electricity, but not enough people care to actually make a different and cognizant effort to decrease consumption. Albrecht believes the smarter solution is to make the houses and buildings decide when to turn appliances on and off. “The building can solve some hard problems using a computer,” Albrecht said.

In order to come up with automated homes, Albrecht first needed to collect data. First, she received aggregate power data from 400 homes across the United States using the main power grid. Second, she and colleagues launched the Smart* Project. The goal of the project is to “get as much data as possible out of a home,” Albrecht said, using inexpensive, off-the-shelf components. Albrecht wired three homes in the Berkshire area with multiple sensors to measure energy usage and consumption. She then gathered incoming and outgoing power information from her house. Additionally, she measured each circuit, light and plug in her house to determine how each appliance used electricity. Albrecht is currently working on the process of “embedding controls in something that’s not meant to have them,” she said. This summer, she plans to work with some students to “hack a coffee pot” in order to monitor its energy usage.

Albrecht’s research involves analyzing all of these data to determine when somebody is home or away. “We want the house to be smart enough to not even have the people living in the house know what’s going on,” Albrecht said. The best application of this research right now is the thermostat, so, with enough data, the thermostat can turn on or off depending upon when people are home or away.

“The more information you have, the easier it is to work with,” Albrecht said. This fact is both beneficial – as it allows the smart home to determine when the homeowner is home or away – and possibly detrimental – as a person could hack the smart home and learn when people are home or away. In addition to working with the power usage data to come up with a smart home that helps save energy, Albrecht is also researching ways to mask the data to prevent somebody from analyzing it and determining if somebody is home.

“Smart buildings need to detect and predict occupancy to make optimal decisions,” Albrecht said. Albrecht plans to continue research ways to increase the accessibility and success of this method to allow homeowners to utilize it in the near future.