Prof. Ningfang Mi on Predicting the Unpredictable

There’s an almost mythical quality to Professor Ningfang Mi’s work in computing resource management and capacity planning.  Her research aims to revolutionize the behind-the-scenes operations of the Internet and cloud computing to increase quality, improve efficiency, and decrease costs. But in a world with billions of web pages and billions of Internet users, this is no simple task.

Professor Mi’s work in predicting the unpredictable may, to the layperson, border on fortune telling, but the truth is somehow more fascinating: she employs a comprehensive and multidisciplinary approach that relies on sophisticated mathematical and statistical modeling. Depending on the particulars of a situation, Mi refines models to represent real systems based on advanced queuing techniques.  Using these techniques, such as Markovian arrival processing, she in turn designs increasingly sensitive and accurate systems for balancing dynamic workloads. The success of her work in this critical field is turning heads; this year, the U.S. Air Force Office of Scientific Research announced that Mi would be one of 42 researchers nationwide to receive a Young Investigator research grant. The Young Investigator project, “Creating An Integrated Management Layer To Administer Heterogeneous Resources in Dynamic Workflow Clusters,” will develop a system that helps heterogeneous hardware in diverse settings work together in well-oiled unison.

Perhaps one of the most interesting aspects of Professor Mi’s research is its emphasis on stunningly simple and elegant goals for exceedingly complex challenges--all of which Mi exudes obvious enthusiasm for. Yet her field is in its infancy, leaving one to wonder how she landed here. Was she somehow adept at predicting the unpredictable on a larger scale and entered computer engineering with the goal of researching a problem that didn’t yet exist?

The answer to that, of course, is no; it was primarily by chance that she began to learn about modeling of heterogeneous cloud-based systems. In graduate school, before cloud computing catapulted to prominence, Mi studied computational geometry.

“It was very theoretical and had lots of calculations,” Mi recalls. She wasn’t sure where her training would lead her, but--crucially--she kept an open and curious mind. Upon entering her PhD program, she took a variety of courses to gain more exposure to different paths. It was in a class taught by her future advisor, Evgenia Smirni, during her second year of her PhD at William & Mary, that Mi started to become passionate about the practical applications of her training in modeling and resource management.

Nearly five years into her career as a faculty member in Electrical and Computer Engineering at Northeastern, Professor Mi seems to be reflecting on a different life when recalling the beginnings of her engineering training. She never could have imagined the kind of fascinating and cutting-edge work she’d be doing today.

But now, she is the advisor to three doctoral and a master’s students in her research group. In the fall 2014 semester she’ll be teaching an undergraduate course, EECE3326: Optimization Methods. Professor Mi relates that students often come to her seeking insider tips on which field to study, wanting to know which area will be a good bet--in a sense, they want help predicting their own futures. Professor Mi, however, provides these students with more open-ended guidance.

“I use my story to tell them that the initial area may be the final area you are interested in or want to work on.” Regardless, though, it’s good to be pretty sure. “You must be interested” in your research field, she says, “because as a PhD student or researcher, it will be not your whole life, but it will be most of it!”  She then describes the typical conversation in which students mention fields they may be weakly interested in as though seeking her blessing.

“I tell them, ‘well, that’s a good area, but maybe there’s some area that will fit you better.’ So definitely explore different things: special topics, research classes; talk with professors with different projects.”

“Then,” she adds, “you will find what fits you best.”

[Last updated 07/2014 by K. Ruben]