Keynote: Challenges in Networking for AIML clusters
Speaker: Prof. Costin Raiciu
University Politehnica of Bucharest
Abstract: Despite the commonly held belief that the network is not the bottleneck in datacenters, training machine-learning workloads is network bound and has lead to the deployment of fully provisioned folded Clos topologies. Even so, in production that communication makes up 15% to 75% of training time depending on the model, and this is expected to become worse as training clusters increase in size; worse, existing transports fail to properly utilize the available core capacity. We analyze in detail how we can improve networking for AI/ML training, with a particular focus on understanding how packet spraying can be used to achieve good performance for difficult collective communications such as all-to-all. We will give an overview of the Ultra Ethernet Consortium, a novel standardisation body that is now working on defining the transport for AIML networks.
Bio: Costin Raiciu is Chief Architect in Broadcom's Core Switching Group and Professor at University Politehnica of Bucharest where he leads the Netsys group. Costin has received his B.Sc. and M.Sc. from University Politehnica of Bucharest in 2003 and 2004, and his PhD from University College London in 2011. Costin's work is at the area of networking, operating systems and verification. His main work includes design and implementation of Multipath TCP [RFC8684, RFC6356], lightweight virtualization (LightVM [SOSP17] and FlexOS [ASPLOS22]) and verification (Symnet[SIGCOMM'16], BF4 [Sigcomm'20]). Costin is very keen on pushing his research work into production, with Multipath TCP a prime example of deployed work. Costin is now part of the team standardizing next generation networking at the Ultra Ethernet Consortium, a protocol that is partly based on his prior work called EQDS (NSDI'22).
Keynote: Unveiling AI-Native Core Network Configurations
Speaker: Prof. Nirwan Ansari
New Jersey Institute of Technology (NJIT)
Abstract: 3GPP sets the stage for a groundbreaking transformation towards AI-Native operations in the forthcoming 6G core networks (CNs). Departing from the disparate and non-standardized use of artificial intelligence (AI) in 5G CNs, 6G CNs herald a new era with a revolutionary redesign, where AI takes center stage as the fundamental driving force behind all network functions – an epochal shift known as AI-Native. Delving into this paradigm shift, our keynote presentation illuminates the pivotal role of AI in reshaping network architecture.
Bio: Nirwan Ansari, Distinguished Professor of Electrical and Computer Engineering at the New Jersey Institute of Technology (NJIT), holds a Ph.D. from Purdue University, an MSEE from the University of Michigan, and a BSEE (summa cum laude with a perfect GPA) from NJIT. He is a Fellow of the Institute of Electrical and Electronics Engineers (IEEE) as well as the National Academy of Inventors (NAI). He authored Green Mobile Networks: A Networking Perspective (Wiley-IEEE, 2017) with T. Han, and co-authored two other books. He has also (co-)authored over 700 technical publications, with more than half of them published in widely cited journals and magazines. He has served as a guest editor for numerous special issues on various emerging topics in communications and networking. Currently, he serves as the Editor-in-Chief of IEEE Wireless Communications and has been on the editorial/advisory board of over ten journals. His current research focuses on green communications and networking, edge computing, drone-assisted networking, and various aspects of broadband networks. He was elected to serve on the IEEE Communications Society (ComSoc) Board of Governors as a member-at-large. He has served as the Director of ComSoc Educational Services Board, chaired various technical and steering committees within ComSoc, and has served on many committees such as the IEEE Fellow Committee. He has actively participated in organizing numerous IEEE International Conferences/Symposia/Workshops. Among his many recognitions are several excellence in teaching awards, multiple best paper awards, the NCE Excellence in Research Award, several ComSoc TC technical recognition awards, the NJ Inventors Hall of Fame Inventor of the Year Award, the Thomas Alva Edison Patent Award, the Purdue University Outstanding Electrical and Computer Engineering Award, the NCE 100 Medal, the NJIT Excellence in Research Prize and Medal, and designation as a COMSOC Distinguished Lecturer. He has also been granted more than 40 U.S. patents.
Keynote: 6G Intelligentization: AI and More-than-Human Natural Intelligence for Exiting the Anthropocene
Speaker: Prof. Martin Maier
Optical Zeitgeist Laboratory - INRS), Montreal, QC, H5A 1K6 Canada
Abstract: 6G will be transformative by advancing network softwarization to network intelligentization, whereby sustainability will be front and center in all development and design decisions throughout the life cycles and value chains of 6G. Some argue that the true value of AI will not be in how it enables society to reduce its energy, water, and land use intensities, but rather, at a higher level, how it facilitates and fosters environmental governance, i.e., the formal and informal rules that govern human behavior relevant for long-term sustainability. In the past, the single biggest enabler of today’s AI progress has been online data produced largely by human users. Future AIs will go offline and explore the expanding universe of Generative AI models, which have barely scratched the surface of embodied data modalities such as biology. The global information transmission rate of the nature-made biosphere is estimated to be ~1024 bits/second, roughly 9 orders of magnitude higher than that of our man-made technosphere. This keynote elaborates on the mutually beneficial symbiosis between advanced digitalization (Generative AI and Metaverse) and biologization (also known as biomimetics) for adapting our species back to nature via human-AI interaction in order to exit the Anthropocene by moving away from anthropocentric thinking, because nothing will change until we do. Toward this end, we explore a new non-anthropocentric kind of AI that differs from today’s anthropocentric, human brain-inspired AI by incorporating the many other ways in which more-than-human natural intelligence operates.
Bio: Martin Maier is a full professor with the Institut National de la Recherche Scientifique (INRS), Montréal, Canada. He was educated at the Technical University of Berlin, Germany, and received MSc and PhD degrees both with distinctions (summa cum laude) in 1998 and 2003, respectively. In 2003, he was a postdoc fellow at the Massachusetts Institute of Technology (MIT), Cambridge, MA. He was a visiting professor at Stanford University, Stanford, CA, 2006 through 2007. He was a co-recipient of the 2009 IEEE Communications Society Best Tutorial Paper Award. Further, he was a Marie Curie IIF Fellow of the European Commission from 2014 through 2015. In 2017, he received the Friedrich Wilhelm Bessel Research Award from the Alexander von Humboldt (AvH) Foundation in recognition of his accomplishments in research on FiWi-enhanced mobile networks. In 2017, he was named one of the three most promising scientists in the category “Contribution to a better society” of the Marie Skłodowska-Curie Actions (MSCA) 2017 Prize Award of the European Commission. In 2019/2020, he held a UC3M-Banco de Santander Excellence Chair at Universidad Carlos III de Madrid (UC3M), Madrid, Spain. Recently, in December 2023, he was awarded with the 2023 Technical Achievement Award of the IEEE Communications Society (ComSoc) Tactile Internet Technical Committee for his contribution on 6G/Next G and the design of Metaverse concepts and architectures as well as the 2023 Outstanding Paper Award of the IEEE Computer Society Bio-Inspired Computing STC for his contribution on the symbiosis between INTERnet and Human BEING (INTERBEING). He is co-author of the book “Toward 6G: A New Era of Convergence” (Wiley-IEEE Press, January 2021) and author of the sequel “6G and Onward to Next G: The Road to the Multiverse” (Wiley-IEEE Press, February 2023).
