Professor Mina Tahmasbi Arashloo has been awarded a new Canada Research Chair (CRC) in Minimizing Human Error in Operating Modern Networks.
She is among nine researchers at the University of Waterloo, awarded $8.9 million in funding to support their research as part of the illustrious CRC program.
In 2000, the federal government launched the CRC Program to position the nation as a global leader in research and development. Every year, it invests $311 million to recruit and retain top talent and to support academic research and training at Canadian post-secondary institutions.
“Congratulations to Mina for receiving this honour,” says Raouf Boutaba, University Professor and Director of the Cheriton School of Computer Science. “She joined our School two years ago and has made significant research strides. Her research approach is highly innovative, applying techniques from formal methods research in computer science to rigorously reason about network performance.”
Professor Arashloo has been appointed the Tier 2 Chair, which honours exceptional emerging researchers. She is being recognized for her research in networked systems, particularly software-defined networking (SDN) and programmable data planes. Her work has enhanced computer networks in a way that could help prevent network errors like low performance, malfunctions, or outages.
“I am excited and very grateful for this award. Computer networks are one of the cornerstones of modern society, and we need to make sure they can operate correctly and efficiently so we can continue relying on them in our day-to-day lives. This generous award enables us to explore how to design new techniques, systems, and frameworks that can help modern networks become more robust and efficient.”
Professor Arashloo joined the University of Waterloo’s Cheriton School of Computer Science as an Assistant Professor in 2022. In 2014, she obtained her Bachelor of Science in Computer Engineering at the Sharif University of Technology in Tehran, Iran. From 2014 to 2019, she attended Princeton University, where she received her Master’s and PhD degrees in Computer Science. She also worked as a postdoctoral researcher at Cornell University.
She is the recipient of various prestigious awards and grants such as the Heidelberg Laureate Forum’s Young Researcher (2023), N2Women’s Rising Stars in Networking and Communication (2021), Cornell University’s Presidential Postdoctoral Fellow (2019-2022), Association for Computing Machinery’s Special Interest Group on Data Communication’s (ACM SIGCOMM) Dissertation Award (2019), Siebel Scholars Award (2019), and Microsoft Research’s Dissertation Grant (2019).
Computer networks help devices, like phones and computers, exchange and share information. For example, when someone Googles something on their phone, their query is transferred to a computer in a Google data centre. That computer will exchange information with other computers in the data centre to compile a list of answers. Afterwards, the response will circle back to the person’s phone. All the data exchanges in this process are achieved through computer networks.
These networks are deeply ingrained in our society—it’s the backbone of everyday online activities like work, banking, shopping, learning, and media access. As a result, it’s important that they work and operate effectively. If not, the resulting disruptions could potentially affect millions of people. One notable example is the 2022 Roger Communications outage, which affected one of Canada’s biggest telecommunications service providers. Although this incident lasted one day, it affected 12 million Canadians. Many of them couldn’t access any critical services like emergency services or banking.
A major source of network problems is human error. Modern networks are extremely complex since they are composed of thousands of devices, each with its own sophisticated software and hardware. On top of that, computer networks are constantly evolving: their components are frequently added, removed, and updated. This complexity is a hotbed for mistakes: network engineers and operators often miss subtle issues in network components that become noticeable in interaction with other network components or when processing live traffic.
With the CRC funding, Professor Arashloo will create techniques, systems, and tools to minimize human error in operating modern networks. Her team will focus on automated analysis, which can catch network issues before deployment. They will also design network programming platforms that can generate correct and efficient implementations for network functionality, and real-time monitoring platforms for network-wide properties.
Previously, these three strategies have been studied separately and only for certain network components or basic properties. What distinguishes Professor Arashloo’s research is that it will combine them in cohesive frameworks that focus on the more advanced properties modern networks are expected to satisfy. This approach will ensure that networks can operate effectively and can help network engineers and operators identify any potential oversights during the design and deployment stages.
Professor Arashloo’s past projects have focused on improving network robustness, efficiency and performance. She has co-developed FPerf that can automatically examine network performance under a wide range of possible data exchange patterns. The number of possible data exchange patterns is vast and constantly evolving, and as such difficult to analyze the resulting network performance for all of them.
FPerf models network components that are important in determining performance, like network queues, in logical terms. It also provides an interface for its users to ask questions about network performance metrics like throughput and queue sizes. By combining several sophisticated techniques such as model checking and syntax-guided synthesis, FPerf can automatically identify the data exchange patterns that lead to performance issues, offering a comprehensive analysis that previous methods could not achieve.
FPerf has attracted interest in the networking community, reflecting the need for automated analysis tools to handle the growing complexity of modern networks. Most of the existing work in this area focused on analyzing the correctness of basic network connectivity. FPerf stands out as the first framework to enable analyzing the performance of data exchanges over the network. The early positive reception of FPerf indicates its potential to drive future research and inspire new methods for analyzing network performance, especially as networks continue to evolve.
Professor Arashloo also collaborated with researchers from Princeton University and the Massachusetts Institute of Technology (MIT) to design Tonic, a hardware architecture that can support network transport algorithms in high-speed network hardware. Transport-layer algorithms help prevent network congestion by controlling how quickly data is sent across the network. They also ensure reliable data delivery by resending any data that might get lost along the way. Researchers and practitioners have implemented individual transport-layer algorithms in network hardware. However, these implementation tasks are challenging and error-prone given the various hardware limitations and the need to operate at high speed. Tonic offers a programmable platform where users can effortlessly program and modify network hardware for various transport-layer algorithms. Its key feature is a set of high-speed modules that are reusable across implementations of many transport algorithms. This simplifies the implementation task of transport protocols to writing a few hundred lines of code. So, users can create efficient implementations of important network algorithms with much lower risk of making mistakes.