Upanzi Network Webinar: Aristide Tanyi-Jong Akem (IMDEA Networks Institute Spain)

January 29, 2025

5:00 p.m. CAT

Virtual

Topic: On-Device Machine Learning for Traffic Classification in Programmable Networks

Speaker: Aristide Tanyi-Jong Akem, Ph.D., research assistant at IMDEA Networks Institute Spain
and CMU-Africa alumnus

Abstract: In recent years, network complexity has grown exponentially to support modern applications, rendering traditional approaches to network management inadequate and necessitating greater automation and flexibility. Machine learning (ML)-based traffic classification models in Software-Defined Networking (SDN) have been pivotal in automating tasks like intrusion detection, quality of service prioritization, and routing optimization. However, running these applications in the control plane often leads to response delays of several milliseconds, unsuitable for ultra-low latency scenarios. The advent of programmable user-plane devices, enabled by languages like P4, has opened new possibilities for performing real-time computations directly within the network. By offloading ML models from control-plane servers to user-plane devices such as switches, we can achieve high-speed, low-latency traffic analysis. Yet, these devices face severe constraints in memory, computational operations, and packet processing capabilities, shifting the focus to deploying pre-trained ML models rather than training them on the devices. Existing solutions for on-device ML deployment, while promising, remain limited in addressing complex and dynamic tasks. This talk presents novel approaches to on-device ML-based traffic classification in programmable networks. Our solutions address the limitations of prior work, enabling high-speed, intelligent network operations even under stringent constraints. We also demonstrate the practical applications of our methods, including network intrusion detection, smart grid security, and encrypted traffic classification, highlighting their potential to redefine modern network intelligence and security.

Bio: Akem is a research assistant at IMDEA Networks Institute in Madrid, Spain, and a recent graduate from the Telematics Engineering PhD program at Universidad Carlos III de Madrid, Spain. His research interests cut across machine learning, network programming, and mobile networking, as well as their applications to network security, IoT, and energy. Akem has developed several solutions for deploying machine learning models into P4-programmable hardware like switches for high-speed inference. He has also made research visits to Orange Labs in France, Ranplan Wireless, and the University of Cambridge, in the United Kingdom. Akem has co-authored multiple peer-reviewed articles, some of which have been presented at top conferences like IEEE INFOCOM. Before his PhD studies, Akem obtained a Master of Science in Electrical and Computer Engineering from Carnegie Mellon University Africa in Rwanda in 2020, and a Master of Engineering in Telecommunications from the University of Yaounde I in Cameroon in 2018.

Register for Upanzi Network Webinar: Aristide Tanyi-Jong Akem (IMDEA Networks Institute Spain)

Upcoming Events

Browse All Upcoming Events