CMU research center aims to improve networks connecting devices to cloud
It is early in the era of the Internet of Things, but it already appears our existing networks can't handle it.
Smart devices get kicked off the network, go offline and stop working.
“I think people are already getting sick of Internet of Things technology in their homes getting flaky,” said Anthony Rowe, an associate professor of electrical and computer engineering at Carnegie Mellon University. “Nobody wants to constantly reboot their TV or their thermostat.”
Rowe heads a new research center at CMU aimed, in part, at fixing the Internet of Things. The Computing on Network Infrastructure for Pervasive Perception, Cognition, and Action Research Center — CONIX for short — will bring together scientists from six U.S. universities to develop a smarter, most robust network that could improve everything from your connected refrigerator and sensors in a smart city to augmented reality and autonomous robots.
The creation of the center was announced this week.
The research center will be based at CMU. It received $27.5 million from Semiconductor Research Corp. and the Defense Advanced Research Projects Agency. It secured another $6.5 million in internal funding.
Rowe said the center's work will focus on the network that connects devices to the cloud. That network now shuttles packets of information from those devices to the cloud. They are computed and analyzed in the cloud, and then sent back to the devices. The center hopes to create a network that can do some of the computing now done in the cloud, Rowe said.
That type of technology could cut down on the time it takes data to go out to the cloud and come back and make networks more reliable and secure, Rowe said. It could help the development of smart cities by taking in data from sensors and being able to process them. It could make augmented reality work more quickly and smoothly by better connecting digital information with the real world. And it could enable swarms of drones to communicate with one another to go on search and rescue missions or inspect infrastructure and beam back data in real time, Rowe said.
Rowe estimates it could take 10 to 15 years to develop and deploy this type of network. The team at CONIX has its work cut out for them.
“We really don't know how to do this yet,” Rowe said. “We don't know how to program these systems.”
The CONIX team includes researchers from CMU; University of California campuses in Berkeley, San Diego and Los Angeles; the University of Washington; and the University of Southern California.