AT&T Labs Fellowship Award Winner
Each year, AT&T Research through its AT&T Labs Fellowship Program (ALFP) offers three-year fellowships to outstanding under-represented minority and women students pursuing PhD studies in computing and communications-related fields. This year three students received fellowships. Yifan Sun is one.
Asking Yifan Sun about her summer project is to be dropped willy-nilly and without preamble into a detailed discussion about phases and frequencies of light pulses, sine and cosine waves, and the interaction of light waves with fiber—subjects most people encounter only in science class. But Yifan is an electrical engineer, and the properties of light and waves are part and parcel of her summer project within the Optical Systems Research Department at AT&T.
From high school (a math and science magnet school in Illinois), she identified electrical engineering as a possible career choice. Of all the avenues open to someone mathematically inclined and with her wide-ranging interests (including carpentry) and extra-curricular activities, electrical engineering appealed to her for its mix of mathematics and programming, predictive modeling, and the hands-on nature of the work.
(If you’re wondering about the connection between fiber optics and electrical engineering, it is this: optical signals have to be converted to electrical signals in order for them to be used in communication systems or to be studied with a computer, which is in the electrical signal domain. Optical computers are being studied but are very far off.)
The title of her project, Frequency and Phase Recovery for High-Order Coherent Modulation, needs explanation (see below), but is best understood by the problem it’s trying to solve: how to continually expand network capacity to keep up with the exponential growth in network traffic?
Building new infrastructure is one way; building faster, higher-capacity networks using the existing infrastructure is another. AT&T is doing both, and has just completed a field trial of a 100-Gigabit/second network (40-Gigabit/sec is the current standard).
Higher-capacity networks are possible today through the use of coherent modulation (cited in the project title), which can detect components within a wave (phase and polarization) and encode different data streams on each component. This opens the possibility to pack more information onto a single wavelength of light and transmit more data on existing fiber bandwidth.
Previously, a direct modulation system could detect only the strength of a signal and was limited to delivering a single data stream.
Coherent modulation works like this: Lasers at a transmitter generate a light beam at a certain frequency; a modulator decomposes the beam for each polarization into its two orthogonal phase components (the sine wave and the cosine wave), encoding a separate data stream onto each wave component. The two data streams ride on the same beam as it travels through an optical fiber to a receiver, which separates the waves and extracts the data.
Right now, two data streams can easily ride one signal, but AT&T is looking to pack much more (hence, the “High-Order” of the project title), but before this can happen, many problems have to be solved. One big problem is that, while in use, vibrations and changes in laser temperatures can cause the frequency to change (causing frequency noise), while intrinsic quantum mechanics can cause the laser phase to change (causing phase noise). If these effects are not perfectly removed, it can be impossible to recover the data at all. It’s harder still when a light beam carries more than one data stream.
Engineers have known about this problem for a long time, and many different solutions exist. One solution is to send known signals and observe the effects when the signal is received. However, this puts constraints on the kinds of signals that can be sent. Yifan is looking at a more predictive model—in effect, blindly guessing the signal’s frequency and phase distortion without knowing if the observed features are caused by phase offset, random noise, or intended data. (This is the phase and frequency recovery part.)
Since she can’t actually see the waves or the phase or the data, Yifan must visualize what is going on inside the optical fibers. She does this using various modeling techniques, including constellation diagrams, which are essentially a shorthand way of representing what happens to a light wave from transmitter to receiver (see sidebar for the in-between steps). This is the part of the task she finds especially rewarding: taking something that in real life is messy, and visualizing it in a very simple, clean, and mathematical way.
Variations in frequency cause the dot, representing the signal, to blur
Yifan’s work, though very experimental at this stage, may someday have direct impact on AT&T networks. Already her mentor Xiang Zhou is impressed and extremely pleased with how quickly her project is progressing. Her contributions are such that she will be named on at least one soon-to-be filed patent, and possibly two.
In the fall, Yifan enters UCLA, and begins studying in depth the fundamental math behind such systems, something that really excites her. However, this was no easy choice, as there were many funding difficulties. For her, the fellowship cinched acceptance in a university hard hit by budget cuts.
So while the fellowship ensures Yifan gets to attend the school of her choice, it also ensures AT&T Research a long-term collaborative relationship with someone already making important contributions, and ones that can support the higher-capacity networks that are needed more and more.
If not electrical engineering, what else?
Biology (medicine), particle physicist, reporter, photographer, film maker, Pixar artist, meteorologist, carpenter. I love engineering, but it’s only part of my passions in life.
Role models in life.
My family and friends, who constantly accomplish amazing feats despite overwhelming obstacles.
Heroes from history.
Einstein, Nelson Mandela, Bill Gates, Walter Bender (from One Laptop per Child), Zhuge Liang from Chinese legends, more if I studied history better.
Academia or areal-world job?
First job in real world, then maybe academia. I do really want to teach at some point, though it doesn’t have to be at a university.
What motivates you?
Being around awesome people who never stop thinking of new things to do.
What single course helped decide your future study?
No single course, though I was influenced a lot by extracurriculars, from science contests in high school to ambitious team projects in college.
Most fun course?
Modeling and Control. It was challenging as a student and mind-wrecking as a TA, but overall a very rewarding experience.
Course you most regretted not taking?
No regrets in life