Full Program Register Now...
Abstract: While Machine Learning has become pervasive as an essential component of many network services and user applications, its energy cost is often difficult to cope with. It is thus critical to improve the sustainability of Machine Learning by reducing its resource demand. This talk tackles this issue while focusing on the emerging approach of Distributed Machine Learning. In particular, we will discuss both the benefits and the challenges posed by Distributed Learning, and the solutions to minimize the energy cost of this approach while fulfilling the performance requirements of a learning process, in terms of learning quality and time. The talk will also discuss Machine Learning model compression as a promising solution to energy saving as well as to the need for the reuse of computing resources. By leveraging model compression, it is indeed possible not only to tune the network and computing resources to the learning requirements, but also to tailor a Machine Learning model around the available resources.
Abstract: In this talk, a forward-looking wireless infrastructure will be presented which includes a new stratospheric access & computing layer composed of HAPS (high altitude platform station) constellations positioned in stratosphere, 20 km above the ground, in addition to the legacy terrestrial layer and the emerging satellite layer. With its bird’s-eye and almost-line-of-sight view of an entire metropolitan area, a HAPS is more than a base station in the air; it is a new architecture paradigm with access, transport, and core network functionalities for integrated connectivity, computing, sensing, positioning, navigation, and surveillance, towards enabling a variety of use-cases in an agile, smart, and sustainable manner for smart cities and societies of the future.
Abstract: Fueled by the availability of more data and computing power, recent breakthroughs in cloud-based machine learning (ML) have transformed every aspect of our lives from face recognition and medical diagnosis to natural language processing. However, classical ML exerts severe demands in terms of energy, memory and computing resources, limiting their adoption for resource constrained edge devices. The new breed of intelligent devices requires a novel paradigm change calling for distributed, low-latency and reliable ML at the wireless network edge. This talk will explore the potential of the Mobile AI paradigm to unlock the full potential of 5G and beyond.
Abstract: To maintain data integrity in the face of network unreliability, systems rely on error-correcting codes. System standardization, such as has been occurring for 5G, is predicated on co-designing these error-correcting codes and, most importantly, their generally complex decoders, into efficient, dedicated and customized chips. In this talk, we show that this assumption is not necessary and is has been leading to significant performance loss. We describe "Guessing Random Additive Noise Decoding," or GRAND, by Duffy, Médard and their research groups, which renders universal, optimal, code-agnostic decoding possible for low to moderate redundancy settings. Moreover, recent work with Yazicigil and her group has demonstrated that such decoding can be implemented with extremely low latency in silicon. GRAND enables a new exploration of codes, in and of themselves, independently of tailored decoders, over a rich family of code designs, including random ones. Surprisingly, even the simplest code constructions, such as those used merely for error checking, match or markedly outperform state of the art codes when optimally decoded with GRAND. Without the need for highly tailored codes and bespoke decoders, we can envisage using GRAND to avoid the issue of limited and sub-optimal code choices that 5G encountered, and instead have an open platform for coding and decoding.
Abstract: Even though 6G is 8 years away, many seem to try pinning down exact features and specifications already today. This leads to some interesting statements, also made by large corporate players. Just one example is the statement that 6G will require a 10x improvement in spectral efficiency while simultaneously achieving at least a 10x improvement in energy efficiency. No theory has yet been known to show how this could be achievable. In the end, operators will need to earn money providing a new level of services at a cost-level which makes these services affordable for mass market consumers. Therefore, here we rather want to ask the question which thrust of improvement could make sense, and why. And then give some possible ways forward. The 4 thrusts for improvements discussed are: trustworthiness, energy efficiency, cost, and new functionality. If we truly believe that 6G will provide an infrastructure for Tactile Internet remote controlled personal mobile robotic and XR applications, we need lower cost, energy efficient, and trustworthy networks that integrate joint communications & sensing. Can this be realistically achieved without infringing physics or theoretic bounds?
Abstract: In order to cope with global challenges humanity is facing in terms of climate change, sustainability and governance, the UN has established the roadmap for years to come on the pillars of the Sustainable Development Goals (SDG). Among the different aspects and specific goals specified within the SDGs, providing resilient and adaptive communication technologies is key towards their achievement. In this presentation, an overview of capabilities and challenges related with communication technologies, with a specific focus on wireless communications will be discussed, with applications related to the scope of implementing context aware environments in Smart Cities and Smart Regions, thus enabling the advancement in several SDGs.