Natasha Devroye is making wireless communications more efficient. The associate professor in UIC’s Department of Electrical and Computer Engineering (ECE) is researching a way to make it reliable, faster, and safer.
An expert in information theory, which is the mathematical theory behind communications, she researches the understanding behind wireless communications like cellphones, Wi-Fi, radio receivers, and more.
“What information theory does is look at the fundamental boundaries of how fast you should be able to communicate,” said Devroye. “It often leads industry to explore different directions. I see it as a guide for industry for what directions to build their systems.”
Consumers and industry experts continuously want technology pushed to the next level as they seek more data and faster speeds. There are more users, and they are streaming video, so they are demanding more bandwidth over wireless devices. Unfortunately, wireless channels are some of the hardest to deal with. Channels like underground fiber optics have extremely high data rates, and this method of communication is quite controlled.
“Transmitting over a wireless medium is much more efficient. You have wireless interference from all the other devices that are transmitting – cell phones, base stations, etc. – so the wireless medium is a hard one to transmit through because of interference. Wireless signals will bounce off something, and you will see reflections, or shadows,” said Devroye, co-director of the Networks Information Communications and Engineering Systems Laboratory at UIC. “You can be in one place and have one bar on your phone, and step over a meter and you have four bars. These are the effects of the wireless channel that are so variable and location-dependent. It’s a hard medium to communicate over, yet, people want wireless communications and more of it. The number of users and the amount of data is continuously increasing. The question is how to we squeeze more data through the same amount of fixed spectra?”
To answer that question, Devroye is researching a way to obtain zero-error capacity, a fundamental performance limit quantifying the number of messages that may be reliably communicated with no error over the primitive relay channel. In this channel, a source communicates with a destination in the presence of an out-of-band relay.
“Understanding how the source and relay may cooperate in communicating messages is interesting theoretically as the relay channel capacity has been open for more than 40 years,” said Devroye. “My research looks at the zero-error capacity, which differs from the more commonly studied small-error capacity, and leads to an unexplored combinatorial problem involving new relaying ‘compression’ graphs.”
Looking at zero-error communication of networks is only one line of her research and it is considered “high risk and high reward” due to it being a very difficult problem. Information theory is a mathematical framework built upon the notion of how fast someone can communicate with vanishingly small (but positive) error in communication. She wants to determine how fast you can communicate with exactly zero error, a stricter requirement.
There are different ways of protecting data from errors. Redundancy is added to the data so even if some of it is corrupted, you can still reconstruct the original data. But it is not a perfect reconstruction, and it may contain some small error.
“What I want to look at is perfect reconstruction or no error in communication,” she said. “It would be a perfect replica of it. What people have traditionally studied is a pretty good replica, but not perfect.”
“Some of the applications in control require a really high reliability of data,” added Devroye. “I want to model this as requiring no errors, to be on the safe side. Making an error in the control of an autonomous car being controlled over a wireless link is a scary thought to me.”
This is why Devroye is researching zero error. Solving the problem with perfect reconstruction will make communication very reliable, and you can do more things through wireless.
Zero error problems use a branch of mathematics called combinatorics. Zero error information theory is considered to be much harder than classical information, which is based on probability theory.
“People are wondering if I will be able to link the two. Whether I will be able to link this really hard math problem and zero error,” she said. “I’m taking a new approach to an old problem. I am after the theoretical limits. The big picture is we want to make the world safer, better, and more capable and flexible.”
Devroye’s research was made possible mainly by five National Science Foundation (NSF) grants including the prestigious NSF CAREER grant. The zero-error effort is supported by a grant entitled “EAGER: Understanding Cooperation Through the Zero-Error Relay Channel.” She is the sole principal investigator for the project. The grant supports exploratory work in its early stages on untested, but potentially transformative research ideas or approaches. The research could be considered especially “high-risk – high-payoff” in the sense that it, involves radically different approaches, applies new expertise, or engages novel disciplinary or interdisciplinary perspectives.
Learn more her research at Devroye Profile.