Cheriton School of Computer Science Professor Raouf Boutaba, his graduate students and international colleagues have received two best paper awards at NetSoft 2019, the 5th IEEE International Conference on Network Softwarization.
Held from June 24 to 28, 2019 in Paris, NetSoft 2019 brought together academia and industry to explore the evolution of software-defined networking, network function virtualization, and cloud networking techniques, which together are driving an unprecedented technological shift in the information and communications sectors.
Best Student Paper Award
The best student paper award at NetSoft 2019 was conferred to Davit Harutyunyan, a visiting PhD student at the Cheriton School of Computer Science in 2018, along with Nashid Shahriar, a PhD student currently supervised by Professor Boutaba, and Davit’s supervisor, Roberto Riggio from FBK CREATE-NET in Italy.
L to R: PhD student Nashid Shahriar and Professor Raouf Boutaba
Their paper is titled Latency-Aware Service Function Chain Placement in 5G Mobile Networks. The fifth generation of mobile communications systems — known as 5G — is on the horizon and has the potential to reshape the mobile communications landscape. 5G systems will make a wide variety of mobile services possible, from machine-to-machine communications and virtual and augmented reality applications to high-definition media delivery and real-time monitoring and control, among many other services.
Such data-intensive applications have stringent quality-of-service requirements. End-to-end latency can be significantly reduced by moving content and computing capability closer to the network’s edge. However, edge nodes have limited computing capacity, so mobile network operators need to decide how to provide computing resources to ensure low end-to-end latency while using network resources efficiently.
“Raouf and his graduate student and colleagues have demonstrated the pros and cons of different service function chain placement strategies by simulating a 5G mobile network empirically,” said Mark Giesbrecht, Director of the David R. Cheriton School of Computer Science. “Their models have the potential to minimize end-to-end latency, lower the cost of service provisioning, and optimize service function migration frequency for 5G mobile networks, crucial hurdles to clear for bringing 5G systems closer to reality.”
Best Paper Award
The best paper award at NetSoft 2019 was conferred to PhD student Shihabur Rahman Chowdhury, recent master’s graduate Anthony Anthony, and master’s students Haibo Bian and Tim Bai, all of whom are supervised by Professor Boutaba.
L to R: Master's student Haibo Bian, Professor Raouf Boutaba, PhD student Shihabur Rahman Chowdhury and master's student Tim Bai
Titled µNF: A Disaggregated Packet Processing Architecture, their research explored the potential of micro network function — which they termed µNF — to help develop and deploy virtual network functions and service function chains efficiently.
Network function virtualization promises to lower operational and hardware costs by moving packet processing from expensive custom hardware to software running on commodity servers. However, the state-of-the-art in network function virtualization has merely replaced the monolithic hardware installations in data centres with monolithic software that performs various network functions virtually.
Although this is a good first step, common packet processing tasks are repeatedly implemented in monolithic virtualized network functions. This stresses the need to re-architect the network function virtualization ecosystem, through modular virtual network function design and flexible service composition.
“Raouf and his students proposed an ingenious solution,” Professor Giesbrecht said. “They demonstrated that a disaggregated packet processing architecture, which they call micro network functions, can be used to deploy virtual network functions and service function chains efficiently. The experimental results show that compared with monolithic software implementations that perform different network functions, their micro network function-based ones achieve the same throughput while using less CPU cycles per packet.”