Vincent Lannurien
d822cdaa6a
All checks were successful
Build and deploy static pages / build-and-push (push) Successful in 19s
113 lines
5.4 KiB
Markdown
113 lines
5.4 KiB
Markdown
---
|
||
title: IC 2024 Digest
|
||
venue: IC
|
||
year: 2024
|
||
date: '2024-01-01'
|
||
tags: []
|
||
paper_count: 10
|
||
draft: false
|
||
---
|
||
|
||
10 papers selected.
|
||
|
||
---
|
||
|
||
### Revisiting Edge AI: Opportunities and Challenges
|
||
|
||
*Tobias Meuser, Lauri Lovén, Monowar Bhuyan, Shishir G. Patil *et al.**
|
||
|
||
**TL;DR** — A multi-author position paper that revisits the state of edge AI, cataloguing deployment barriers and open research problems across hardware, networking, and software layers.
|
||
|
||
**Why notable** — Brings together 19 leading researchers to synthesize the field's most pressing edge AI challenges, making it an authoritative reference for practitioners and researchers planning edge deployments.
|
||
|
||
---
|
||
|
||
### On Causality in Distributed Continuum Systems
|
||
|
||
*Víctor Casamayor-Pujol, Boris Sedlak, Praveen Kumar Donta, Schahram Dustdar*
|
||
|
||
**TL;DR** — Formalizes causal reasoning across cloud-to-edge continuum systems, providing a conceptual framework for tracking cause-and-effect relationships in highly distributed deployments.
|
||
|
||
**Why notable** — Addresses a foundational gap in distributed systems theory that becomes critical when debugging or optimizing multi-tier edge–cloud pipelines.
|
||
|
||
---
|
||
|
||
### Beyond Von Neumann in the Computing Continuum: Architectures, Applications, and Future Directions
|
||
|
||
*Dragi Kimovski, Nishant Saurabh, Matthijs Jansen, Atakan Aral *et al.**
|
||
|
||
**TL;DR** — Surveys non-von Neumann architectural paradigms—neuromorphic, in-memory, and dataflow computing—and maps them onto continuum computing use cases spanning edge to cloud.
|
||
|
||
**Why notable** — Offers a rare cross-cutting view of how emerging hardware architectures reshape the design space for distributed Internet applications.
|
||
|
||
---
|
||
|
||
### ARASEC: Adaptive Resource Allocation and Model Training for Serverless Edge-Cloud Computing
|
||
|
||
*Dewant Katare, Eduard Marin, Nicolas Kourtellis, Marijn Janssen *et al.**
|
||
|
||
**TL;DR** — Proposes ARASEC, a system that jointly optimizes resource allocation and on-device model training for serverless functions deployed across edge and cloud nodes.
|
||
|
||
**Why notable** — Demonstrates measurable efficiency gains in a realistic serverless edge-cloud setting, directly informing how operators should provision heterogeneous serverless infrastructure.
|
||
|
||
---
|
||
|
||
### WebAssembly at the Edge: Benchmarking a Serverless Platform for Private Edge Cloud Systems
|
||
|
||
*Giuseppe De Palma, Saverio Giallorenzo, Jacopo Mauro, Matteo Trentin *et al.**
|
||
|
||
**TL;DR** — Benchmarks a WebAssembly-based serverless runtime on private edge cloud hardware, measuring cold-start latency, throughput, and isolation overhead compared to container-based alternatives.
|
||
|
||
**Why notable** — Provides concrete empirical data that practitioners need when evaluating WebAssembly as a lightweight alternative to Docker for edge serverless deployments.
|
||
|
||
---
|
||
|
||
### HeROsim: An Allocation and Scheduling Simulator for Evaluating Serverless Orchestration Policies
|
||
|
||
*Vincent Lannurien, Laurent d'Orazio, Olivier Barais, Stéphane Paquelet *et al.**
|
||
|
||
**TL;DR** — Introduces HeROsim, an open simulator that models serverless function placement and scheduling policies across heterogeneous infrastructure, enabling fair policy comparison without live cluster costs.
|
||
|
||
**Why notable** — Fills a practical tooling gap for researchers and platform engineers who need reproducible evaluation environments for serverless orchestration algorithms.
|
||
|
||
---
|
||
|
||
### Hierarchical Network Data Analytics Framework for 6G Network Automation: Design and Implementation
|
||
|
||
*Youbin Jeon, Sangheon Pack*
|
||
|
||
**TL;DR** — Designs and implements a hierarchical analytics framework that aggregates network telemetry at multiple granularities to automate management decisions in 6G deployments.
|
||
|
||
**Why notable** — Bridges the gap between 6G vision and practical automation by providing a concrete architecture with implementation details and empirical evaluation.
|
||
|
||
---
|
||
|
||
### Digital-Twin-Driven End-to-End Network Slicing Toward 6G
|
||
|
||
*Mahnoor Yaqoob, Ramona Trestian, Mallik Tatipamula, Huan Xuan Nguyen*
|
||
|
||
**TL;DR** — Proposes a digital-twin framework that continuously models and reconfigures end-to-end network slices, enabling dynamic SLA enforcement across heterogeneous 6G infrastructure.
|
||
|
||
**Why notable** — Connects digital twin technology to the operational problem of network slice management, a key requirement for 6G service assurance.
|
||
|
||
---
|
||
|
||
### The Internet of Things in the Era of Generative AI: Vision and Challenges
|
||
|
||
*Xin Wang 0120, Zhongwei Wan, Arvin Hekmati, Mingyu Zong *et al.**
|
||
|
||
**TL;DR** — Examines how generative AI models can be integrated into IoT pipelines for data synthesis, anomaly detection, and on-device inference, and identifies the key resource and privacy constraints.
|
||
|
||
**Why notable** — Provides a structured research agenda for one of the most active intersections in Internet computing, relevant to both IoT platform designers and ML practitioners.
|
||
|
||
---
|
||
|
||
### Distributed Federated Deep Learning in Clustered Internet of Things Wireless Networks With Data Similarity-Based Client Participation
|
||
|
||
*Evangelia Fragkou 0001, Eleftheria Chini, Maria Papadopoulou 0008, Dimitrios K. Papakostas *et al.**
|
||
|
||
**TL;DR** — Proposes a clustered federated learning scheme for wireless IoT networks that selects participating clients based on data similarity, reducing communication overhead and improving model convergence.
|
||
|
||
**Why notable** — Addresses a core practical challenge in IoT federated learning—heterogeneous and non-IID data—with an empirically validated participation strategy.
|
||
|