UOttawa engineers, T.O. tech firm putting AI to work to unclog network traffic jams

U of O stock photo

Researchers at the University of Ottawa are teaming up with a southern Ontario software company on a groundbreaking project that could make dropped Zoom calls and wireless network crashes a thing of the past for remote workers.

Under the four-year agreement announced this week, uOttawa software engineers will partner with experts from Toronto’s RabbitRun Technologies to develop solutions that aim to use artificial intelligence to make software-defined networks – programmable networks that control the flow of information in the cloud – more resistant to events such as cyber attacks, power outages and congested data paths.

The dramatic spike in remote work during the pandemic has put additional strains on cloud-based networks that are dealing with far more traffic than before, explains RabbitRun founder and chief technology officer Pat Saavedra.

That’s made software-defined networks serving home offices more susceptible than ever to security breaches and outages, he adds.

“Managed service providers used to deal with hundreds of networks, and now that all the workers are (at) home, they’re into thousands upon thousands of networks to manage and support,” Saavedra says. 

“There’s a lot more being done at home now – almost everything that you normally did at the office. As a result, these networks are now getting taxed with more traffic.”

Focus on smaller customers

Up to now, Saavedra says, most of the R&D on boosting the performance of software-defined networks has been geared towards enterprise-level systems that serve large customers. 

Now, the engineers at RabbitRun – which focuses on optimizing networks for SMEs and remote workers – are hoping to change that, with additional brain power courtesy of uOttawa.

The T.O. firm’s experts will feed data to a team of researchers in the capital led by associate professor Mehrdad Sabetzadeh and his colleague Shiva Nejati. 

The uOttawa team will try to detect anomalies in network data-flow patterns to predict where traffic jams are most likely to occur and when potential security breaches are looming on the horizon. They will then reprogram the networks to anticipate such events and “steer clear” of those obstacles and threats.

"It’s much better to have a concrete solution to one problem as opposed to an abstract solution that doesn’t work for anything."

Sabetzadeh says this project differs from much of traditional academic research, which is generally based in theory and is not always driven by a need to solve real-world problems.

“It’s much better to have a concrete solution to one problem as opposed to an abstract solution that doesn’t work for anything,” he explains.

The veteran software engineer says the partnership will give uOttawa students valuable experience with cutting-edge technology while providing RabbitRun with a cost-effective R&D effort. While the Toronto firm is contributing cash and equipment to the project, the investment is nowhere near as steep as a full-scale commercial research program would cost.

“Eventually, we hope that we will build usable prototypes that are good enough that the partner can take and turn into commercial products,” Sabetzadeh says. 

“We take the risk, and before RabbitRun spends a couple of million dollars in rolling out this technology, they have a very good idea about feasibility.”

Saavedra, who was introduced to Sabetzadeh and his team through the University of Ottawa’s business development office, believes the research effort will pay long-term dividends for SMEs and remote workers. 

“The more success we have, the more impact it will have opening up those (home) markets – allowing people to do what they’ve got to do under these circumstances and communicate, collaborate and just work,” he says.

“We’re already seeing areas where we can dig into right now and have something viable for the markets in short order.”