|Two labs, two high-impact missions|
By Abby Cohn
BIG DATA, BIG CHALLENGES: What to do about the massive aggregation of digital data? For researchers at the AMPLab, the challenge is not just finding ways to store it all, but also developing tools that will efficiently manage and analyze this information, thus maximizing its value. LUIS M. MOLINA
Two new research ventures at Berkeley Engineering have boundary-shattering visions for the future of computing.
Meet the Swarm Lab and AMPLab.
Jointly unveiled at the recent Berkeley EECS Annual Research Symposium (BEARS), the labs have distinct missions. The Swarm Lab will advance work in tiny wireless sensors capable of linking our homes, cities and bodies to the cyber world. The AMPLab will focus on solutions to the growing challenge of storing, accessing and analyzing a deluge of data that has begun overwhelming today’s technology.
Each is assembling an interdisciplinary team composed of leading computer scientists from multiple specialties.
“You really need expertise across a wide spectrum of the computer landscape,” says Michael J. Franklin, professor of computer science and director of the AMPLab, whose initials stand for algorithms, machines and people.
Housed in Soda Hall, the AMPLab is creating cutting-edge tools to manage and extract valuable information from the plethora of data being collected digitally. Its mission comes in response to the phenomenon known as “big data” or the “data deluge”: Modern computing is gathering so much information—ranging from online mouse clicks to human genome sequencing to telescope imagery of the universe—that it lacks the ability to store it, let alone analyze it to understand trends or make predictions. “It used to be that you had to be a big phone company or bank to have a data problem,” says Franklin. “Now everyone is starting to collect more data than they can make sense of.”
The AMPLab is tackling those challenges by taking a three-pronged approach. Researchers plan to improve the efficiency and quality of machine learning (algorithms), scale up datacenters (machines) and leverage the input of human intelligence and activities through crowdsourcing (people). Instead of advancing each dimension independently, the lab will work to integrate the three into a unified system. “We have world-class people in all dimensions,” Franklin says. “We’re taking a holistic view of the system architecture.”
Illustrating the value that can be mined from massive collections of information, the AMPLab is already teaming with several real-world projects using big data to create personalized genomics information, crowd-based public opinion forums and urban development simulations. Another early participant is the Mobile Millennium project. Headed by Alex Bayen, associate professor of civil and environmental engineering, that endeavor is creating a real-time traffic monitoring system by processing millions of data points daily from GPS-enabled cell phones and road sensors.
The AMPLab is supported by Google, SAP and some 10 other corporate sponsors. The lab is expected to run for five years.
ABUZZ WITH POSSIBLILITIES: Among other projects, researchers at the Swarm Lab will develop flexible, paper-like and wearable materials from innovative components, such as a UC Berkeley research team’s artificial e-skin made of nanowires, pictured above. COURTESY ALI JAVEY AND KUNIHARU TAKEI
Meanwhile, construction of the Swarm Lab, which will occupy the entire fourth floor of Cory Hall, is expected to start soon. Launched with major support from Qualcomm Inc., the lab will explore and develop smart sensor networks that can be embedded in walls, streets and even the human body. These trillions of tiny, wireless sensors—collectively known as “the swarm”—will capture information about ourselves and our world and provide new ways of interacting, says Jan Rabaey, professor of electrical engineering and computer sciences.
He hopes to open the Swarm Lab by the end of this year.
The lab will serve as an incubator for swarm applications and platforms. Potential applications for the technology include systems that monitor environmental conditions, energy use and personal health. Down the road, swarms could augment reality by creating 3D simulations complete with such sensory experiences as touch, sound and smell.
If, for instance, a harmful chemical starts leaking from a local oil refinery, the wireless network of the future could notify nearby residents with a message that pops up on their bathroom mirrors. Or sensors in the human body might provide instant cholesterol readings or constant EKG monitorings for patients with heart problems. In the future of e-commerce, shoppers could try on clothes virtually and get sensory feedback if the shoes they’re considering will pinch their toes.
Rabaey believes ubiquitous sensory swarms will become a reality within the next 20 years as today’s pad devices “disappear and fade away in the walls or our body.” The biggest hurdle to that vision is the anticipated public concern about privacy, health and other issues related to embedded sensors, he says. Rabaey is confident those worries can be overcome with such measures as careful privacy controls and public education.
He equates the potential of sensory swarms to that of mobile devices when they first debuted. “I believe it’s really going to revolutionize how we interact with the world and how we’re going to interact with each other,” he says.