A Look Inside Google and Carnegie-Mellon's IoT Campus

For the "Internet of Things" to thrive, all it needs is for all devices to get along—which is currently wishful thinking. Last week, however, Google announced a partnership with Carnegie-Mellon University, which is leading a collaboration of faculty from several other academic institutions on a project to jumpstart the Internet of Things revolution. Their plan: Build a universal platform that lets any device talk to any other device. And fittingly, that master-key solution will be open source.

The problem is that IoT software and devices are mostly proprietary, built by each company and working well within their own sandboxes, but they don't communicate well together. The joint project between CMU, Cornell, Stanford, Illinois at Urbana-Champaign, and Google wants to wipe away the private-industry middlemen that keep sensors in separate sandboxes by creating a new, open platform: GIoTTO.

The CMU team leader, Anind Dey, and his fellow professors are building GIoTTO's middleware, which is a pretty technical job—stitching all the right software together so any sensor you pick up off the shelf (measuring temperature, pressure, light, etc.) will work with your system. Dey wants their platform to receive info via any signal type—whether BLE, passive infrared, or otherwise—and show people that info in ways they'll understand.

To do that, Dey and his team will make a lot of proof-of-concept examples to test their tech, building sensor networks on campus to make the campus a "living laboratory" for people to discover IoT and come up with ways to apply the world of IoT to their field of study. The expansion across campus (and potentially out into the city of Pittsburgh) has three goals: Discover different use cases as diverse people outside Dey's labs play with the IoT sensors, drive iterative development of GIoTTO's open infrastructure by adapting it to those different use cases, and allow more people to experience the Internet of Things and study how they use it.

Dey's team will also be teaching undergraduate and graduate courses on the future of IoT (and maybe plucking choice ideas from crafty students). The courses aren't just in computer science: Dey's team is a collection of seven computer science and engineering professors (and their postdocs and graduate students, bringing the team size up to about 50ish people) specializing in hardware, machine learning, and human-computer interaction. But because CMU is a democratic campus—very few things are pushed on faculty from on high—Dey's team can't just spread sensors willy-nilly on other departments' turf. The plan: Publicly display some sample end-user applications that demonstrate what you can do with sensor-collected data, and wait for word of mouth to spread. Eventually, Dey hopes, other departments will come asking Dey's team for a sensor cluster of their own.

"There will be a ton of different objects going into our infrastructure—so we want to be able to query them. We want to be able to ask any sensor and pull or subscribe and tell me when something interesting is happening," says Dey. "Our idea is that I as an end user do not understand what readings of barometric milligrams of mercury mean. What I want to be able to do is take a barometer and put it on my wall, open and close my window 10 times, and not get a response in terms of pressure—instead, I have a sensor that tells me whether my window is open or closed."

The sensors themselves measure basic environmental information—temperature, proximity, barometric pressure, seismic or vibration activity, light—but Dey's team will be able to draw activity conclusions from multiple readings. For example, the proximity sensors can track how many people are in an area or in line for an event, while a temperature sensor could tell the team if someone switched on a new pot of coffee. The type of sensor isn't terribly important to the experiment, says Dey: His team is really just excited to hear what other departments want to track, and Dey's team will finagle a way to track it using available sensors. With such a liberal and technologically aware population, Dey is sure there will be uneasiness about data tracking. Until their project's data collection becomes a norm, Dey's team will put up signage to indicate what data the sensors are collecting and where it is being collected.

In a year, Dey hopes to move on to phase two in his diabolical plan: Spread those sensors across the city of Pittsburgh to continue collecting sensor data—but also to satisfy his inner transportation wonk. Dey would like to put a sensor network all over Pittsburgh's bridges, which are crucial infrastructure for a city spread over three rivers, and use sensors to dig into the city's bus network. Get enough data, and Dey could advise the local transit authority to change the timing of this bus line or move that line three blocks over or even eliminate a line and replace it elsewhere, says Dey.

Half a dozen of Dey's professor peers had already been independently working on IoT applications when Google came calling for proposals for what became the GIoTTO project. Dey and his peers were picked to handle the middleware of the Google-helmed endeavor to build the new GIoTTO platform, but other academic teams are handling different pieces of the GIoTTO project pie: Another CMU team is handling privacy, a team from the University of Illinois at Urbana-Champaign is handling security and networking protocols, and the team at Cornell University is focusing on keeping all that IoT sensor data secure and stopping intrusive folks from siphoning all that IoT data.

Why all the focus on security and privacy? Right now, a hacker prying into someone's online world will find info from online banking and other accounts, but IoT devices provide a lot of behavioral data about where you go and what you do. And since the GIoTTO platform is designed to work with off-the-shelf sensors, hackers have had plenty of time to find ways inside that tech.

"It's an incredibly hard problem to solve. I can't think of a single device on our network that hasn't been hacked in the past, whether it's an OnStar car or an insulin pump," says Dey.

And of course, having Google onboard means Dey's team might get to play with Google's toys. Dey's team is working to get access to the NEST family of devices and hopes to get access to become developers for Google Now. The scope of the project is large—to usher in the age of IoT adoption by building a platform that works with any sensor—but Google and the academics are committed to keeping the whole thing open-sourced. They hope to have the first version of GIoTTO out by mid-fall, when the various academic teams meet in late October, says Dey.

"We're really trying to make it as open as possible. As the IoT ocean gets bigger, all boats rise," says Dey.