DRUID-NET’s organized the Intelligent Distribution of Computing in Cloud Continuum (ID3C) Workshop at Meditcom2022 from 5th-8th September at Athens, Greece. Online
Pictures from the Event
Background and Motivation
The proliferation of IoT solutions is driving the development of novel computing platforms that cope with the limitations of sensor/actuation devices and mobile devices, by offloading computing complexity onto the network. As a result, new computing paradigms that support diverse applications’ needs have arisen including cloud, fog and edge computing. Increasingly hybrid approaches are being adopted to provide performance trade-offs among those distribution models according to changing network conditions and application requirements. This trend is foreseen to continue to grow especially in smart environments powered by post-5G networks. Processing will have to be delegated via novel intelligent coordination strategies over dynamic networks, including cloud, fog and edge elements. There is a need for ubiquitous, context-aware, robust solutions that dynamically orchestrate computing tasks among these models.
Session 1 (9:00 – 10:30) – “Distribution of Computing in Cloud Continuum” Papers Session
Computational Offloading for the Industrial Internet of Things: A Performance Analysis, by Sirine Bouhoula, Marios Avgeris, Aris Leivadeas, Ioannis Lambadaris
Online Learning for Industrial IoT: The Online Convex Optimization Perspective, by Livia Elena Chatzieleftheriou, Chen-Feng Liu, Ioardanis Koutsopoulos, Mehdi Bennis, Mérouane Debbah
A Delay-Aware Approach for Distributed Embedding Towards Cross-Slice Communication, by Ioannis Dimolitsas, Dimitrios Spatharakis, Dimitrios Dechouniotis, Symeon Papavassiliou
Energy Efficient Placement of ML-Based Services in IoT Networks, by Mohammed Moawad Alenazi, Barzan Yosuf, Sanaa Hamid Mohamed, Taisir El-Gorashi, Jaafar Elmirghani
Session 2 (11:00 – 12:30) – “Smart Distribution of Computing in. Dynamic Networks” Invited Talks Session
Learning and resource allocation problems in future wireless edge architectures (LeadingEdge project), by Iordanis Koutsopoulos (Athens University of Economics and Business, Greece)
Abstract: Future wireless edge architectures will be endowed with AI capabilities so as to respond to unpredictable changes and uncertain network and other conditions. In this talk, we will discuss some concrete resource allocation problems and the interesting new twists they obtain if the assumption of a priori knowledge of network state before decision-making is relaxed. We will also discuss the role of Federated Learning in this context.
Reshaping the Network Infrastructure Towards Massively Scalable Computing (SCORING project), by Prof. Halima ElBiaze (Université du Québec à Montréal, Canada)
Abstract: Current network infrastructure falls short in supporting a massively scalable computing landscape required by most of 5/6G use cases. In-Network Computing (INC) has the potential to shape the Next-Generation Networking Infrastructure (NGNI) into an integrated computation, caching and communication (3C) infrastructure that is needed to fulfill the stringent requirements of emerging applications. The enhancement of 3C integration throughout Cloud-Edge-Mist Continuum provides even further advantages to achieve this goal.
Communication-Aware Dynamic Edge Computing (CONNECT project), by Sinem Coleri (Koc University, Turkey)
Abstract: Many specialized machine learning (ML) algorithms have been developed to learn from sensor measurements, but these assume a centralized setting, where data is available at a central processor with powerful computation capabilities. This centralized approach assumes that the massive amount of sensor data is transmitted to a cloud center, which may not be feasible due to limitations of the devices and channels, not meet the stringent delay constraints of most applications, e.g., controlling an autonomous vehicle, or the privacy requirements of users. To address this problem, we develop real edge intelligence by enabling edge nodes to make local decisions rapidly and reliably in a collaborative manner. This is achieved by developing novel caching, distributed computing and networking methodologies to enable federated/distributed learning taking into account the network dynamics and physical channel variations.
Edge Resource Allocation: The Control Co-design Perspective (DRUID-NET Project), by Dimitrios Dechouniotis (National Technical University of Athens, Greece)
Abstract: In edge computing, resource allocation strategy must take into account both application requirements and resource availability. Towards optimization of communications, control and computing (3C), the emerging challenge of the next-generation industrial controllers is to integrate the resource allocation solution in the design of the control scheme. In this context, we will discuss how we can co-design the feedback controller, ther resource allocation strategy and the task offloading decision.
Distributed Stream Processing on Fog and Edge Systems via Transprecise Computing (DiPET project), by Hans Vandierendonck (Queen’s University Belfast, Northern Ireland)
Abstract: The DiPET project investigates models and techniques that enable distributed stream processing applications to seamlessly span and redistribute across fog and edge computing systems. This talk presents the DiPET approach to distributed scheduling based on transprecise computing, which is built on the observation that computation need not always be exact. This way, we propose a disciplined trade-off of precision against accuracy, which impacts on computational effort, energy efficiency, memory usage and communication bandwidth and latency.
