most significant bits
newsletter of uw computer science & engineering
volume 24, number 2, winter 2015

university of washington
CSE logo
 home CSE Home    MSB archive Winter 2015 MSB    MSB Archive   Contact Info Contact Info 
contents
Internet of Things Research highlights Age progression software Brain-to-brain communication Chair’s message Alumni profile: Captricity 2014 faculty additions Faculty awards and honors TR35 winners Anderson's USENIX awards Domingos' KDD Award Fox IEEE Fellow News and events Taskar Center launches Upcoming events Datagrams
msb24.2 PDF

Make an impact: support CSE

Supporting UW CSE is a left-brain decision. Your gift provides the department with resources for scholarships, fellowships, research support, and funds to build the CSE community we hope you value. Small gifts might not seem significant by themselves, but when pooled together they make great things possible. As you think about giving, please consider making a gift to CSE. Go to our website (www.cs.washington.edu) and select “Support CSE.” You will find a variety of funds that can benefit from your support.

About MSB

MSB is a twice-yearly publication of UW CSE supported by the Industry Affiliates Program.

Editor: Kay Beck-Benton.
Contributors: Ed Lazowska, Hank Levy, Sandy Marvinney, Kristin Osborne, S. Morris Rose
Photo credits: Bruce Hemingway, Mary Levin, Jose Mandojana

We want to hear from you!

Do you have news you’d like to share with the CSE community? Comments or suggestions for future issues of MSB? Let us know! Email the editors at msb@cs.washington.edu and be sure to visit us online at:

Sign up for MSB email

MSB is now available via email. To sign up, send an email to

Leading the charge to power the Internet of Things

Ambient backscatter
Ambient backscatter transforms existing wireless signals
into both a source of power and a communication medium.
It enables two battery-free devices to communicate by
backscattering existing wireless signals.

When Shyam Gollakota arrived from MIT in 2012, the University of Washington gained a rising star in networks and wireless research. Fast forward nearly two years, and the campus -- not to mention the entire computer science and engineering community -- is abuzz over what Gollakota and a group of talented faculty and students have achieved in the quest to power the next great computing revolution: the Internet of Things.

Who needs batteries when you have backscatter?

We will only be able to maximize the opportunity presented by the Internet of Things by developing an unobtrusive and inexpensive means of powering devices while enabling their connectivity.

Ambient backscatter is a system that accomplishes both, taking advantage of existing radio frequency (RF) signals present through cellular and television transmissions. Devices equipped with the technology are able to communicate wirelessly and battery-free by “backscattering” the ambient RF signals, thus avoiding the expense of generating their own radio waves or requiring a dedicated power infrastructure.

Researchers tested prototype devices in a variety of real-world settings, including inside an apartment building, on a street corner, and on the top level of a parking garage. Distances from the nearest television tower -- the source of the signal to be backscattered -- ranged from under half a mile to roughly 6 1/2 miles. Regardless of the distance from the tower, the receiving device was able to pick up the signal sent by the transmitting device. The rate of transmission reached one kilobit per second between devices placed 2 1/2 feet apart outdoors, and 1 1/2 feet apart indoors.

The technology, which was developed with support from a Google Faculty Research Award and the National Science Foundation’s Research Center for Sensorimotor Neural Engineering at UW, earned the Best Paper Award at SIGCOMM 2013.

“We can form a network of our devices out of thin air,” said Joshua Smith, an associate professor of computer science & engineering and electrical engineering. “By repurposing the signals that are already around us to serve as both a power source and a communication medium, we can create self-sustaining systems that require little to no human intervention. The possible applications are endless.”

The team, including graduate students Bryce Kellogg and Aaron Parks, went a step further with the development of Wi-Fi backscatter. With this technology, devices use RF signals as a power source and reuse existing Wi-Fi signals for Internet connectivity.

Wi-Fi backscatter is the first system that has proven capable of connecting battery-free devices to Wi-Fi infrastructure -- that next step needed to truly enable the Internet of Things. Using off-the-shelf Wi-Fi devices, the team was able to demonstrate communication rates of one kilobit per second at a distance just over two meters.

“If the Internet of Things is really going to take off, we must provide connectivity to potentially billions of battery-free devices that can be embedded in everyday objects,” said Gollakota. “By developing the ability to enable Wi-Fi connectivity for devices while consuming orders of magnitude less power than what Wi-Fi typically requires, we are closer to making the Internet of Things a reality.”

Wi-Fi backscatter was presented at SIGCOMM 2014 and made possible with support from the Qualcomm Innovation Fellowship, Washington Research Foundation, National Science Foundation, University of Washington and UW Commercialization Gap Fund.

The ultimate remote control

Another way UW faculty and students are moving us closer toward the Internet of Things is by developing new gesture recognition technology. Two systems, AllSee and WiSee, are attracting attention for defying conventional limitations to open up exciting new frontiers in wireless sensing and human-device interaction.

AllSee is a system that enables gesture control over electronic devices even while they are hidden from sight. The system uses television signals as both a power source and the means by which it receives gesture commands.

“AllSee is the first gesture recognition system that can operate on a range of devices, including those with no batteries, for less than a dollar,” said Gollakota. “It also offers intriguing possibilities for controlling devices too small to incorporate a traditional keypad.”

AllSee
AllSee, bringing gesture recognition
to all devices

Unlike existing technology for gesture control -- for example, that incorporated into the Samsung Galaxy S4 smartphone -- AllSee consumes a miniscule amount of power, enabling it to be always on, and does not require the user to manually initiate its operation. With AllSee, a person could use a gesture to mute his or her smartphone without removing it from a pocket or handbag.

The research team, which includes Kellogg and fellow graduate student Vamsi Talla, tested its prototype using eight different hand gestures. AllSee correctly identified the movements more than 90 percent of the time, from a distance of more than two feet.

Another system, WiSee, moves us closer to a truly smart home by leveraging wireless networks to enable whole-home gesture recognition and control without requiring cameras or instrumentation of the user. Instead, it relies on an adapted Wi-Fi router, which functions as a receiver, and a handful of strategically placed wireless devices.

The system is tuned to detect small changes in the frequency of the wireless signal, called the Doppler shift, caused by a person’s movement. WiSee reads the unique Doppler pattern produced by specific movements and classifies them using a binary pattern matching algorithm.

Using MIMO (multiple input, multiple output) technology, the system is able to lock onto a particular user among a group of people when that individual performs a specific, repetitive sequence of gestures that functions like a password. WiSee then enables that user to use a specific gesture to control a corresponding device, such as turning an appliance on and off.

WiSee
WiSee's gesture control will be signaled by a startup sequence
of gestures such as making a circle with your arm to
control the system before you send the real gesture commands.

“By repurposing wireless signals, we eliminate the need for additional sensors or cameras,” explained Gollakota. “And because Wi-Fi signals can travel through walls, our technology does not require a person to be in the same room as the device that he or she wants to operate.”

This last point sets the system apart from existing gesture-recognition technology familiar to consumers, like Microsoft’s Xbox Kinect. WiSee is simpler, less expensive, and not bound by line-of-sight requirements or sound restrictions. In a user study, its rate of accuracy in identifying nine distinct, wholebody gestures was 94 percent.

WiSee, which was initially funded by CSE, won the Best Paper Award at MOBICOM 2013. The research team is now working on enabling WiSee users to control multiple devices at the same time.

Sensing an opportunity in healthcare

CSE researchers are looking at ways to apply similar principles in healthcare settings. For example, researchers in Gollakota’s lab turned a smartphone into an active system for diagnosing sleep apnea. Called ApneaApp, the system transmits signals on a smartphone and monitors patients’ chest and abdominal movements in sleep, identifying sleep apnea events based on the minute changes caused to the signal reflections from the human body.

ApneaApp demo
Nandakumar demonstrates how
the ApneaApp diagnoses apnea
without using traditional sensors.

ApneaApp offers an alternative to the traditional -- and expensive -- clinical polysomnography test, which requires the patient to spend the night in a sleep laboratory. Unlike other portable alternatives that require instrumentation of the bed or patient and a trained technician to set up, ApneaApp is a truly wireless and contactless tool that can operate on an off-the-shelf Android smartphone.

Working with Dr. Nathaniel Watson at the UW Medicine Sleep Center, Gollakota and graduate student Rajalakshmi Nandakumar were able to demonstrate how the ApneaApp accurately diagnoses three types of apnea: central (when the subject holds his or her breath), obstructive (a complete or partial blockage of the subject’s airway), and hypopnea (when the subject’s breathing becomes shallow). ApneaApp is accurate at distances of up to a meter from the patient, even when he or she is under a blanket.

“There is an incredible opportunity to transform healthcare by looking at how technology can improve diagnosis and outcomes. ApneaApp can potentially transform sleep diagnosis, achieving fine-grained diagnosis -- similar to that achieved in a hospital overnight -- using just a smartphone,” said Gollakota.

Collaboration is the key

UW CSE’s formidable expertise in wireless and sensing research was developed in partnership with our neighbors in Electrical Engineering. Through labs like Gollakota’s that engage students from both departments in hands-on research and joint faculty appointments like Smith, CSE and EE have seized upon a tremendous area of opportunity for the university and for our region. And, as evidenced by the ApneaApp project, the opportunity to collaborate on groundbreaking computer science research extends beyond these two departments on campus.

“By tearing down the barriers between hardware and software, and computer science and other disciplines, we are enabling novel ways of computation, communication and sensing that would not be possible if researchers work in their siloed fields,” said Gollakota. “Computer science affect so many aspects of human life, and working as a member of the CSE faculty offers opportunities to work in a variety of different domains.”

“This is what we do at UW: we collaborate to push the boundaries of our field,” said CSE department chair Hank Levy. “With our ability to attract the best and brightest researchers and our entrepreneurial spirit, we are leading the charge to build systems that have an impact.”

Gollakota, who received the prestigious TR35 award this fall, is focusing on the future. He and Smith launched a startup company, Jeeva Wireless, to commercialize their research with the tagline, “Power and Internet connectivity for the next billion devices.”

“There are so many important problems to solve,” Gollakota observed, “and so little time!”

CSE logo
Computer Science & Engineering Box 352350, University of Washington Seattle, WA 98195-2350 Privacy policy and terms of use