We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.NI

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Networking and Internet Architecture

Title: Interplay between Distributed AI Workflow and URLLC

Abstract: Distributed artificial intelligence (AI) has recently accomplished tremendous breakthroughs in various communication services, ranging from fault-tolerant factory automation to smart cities. When distributed learning is run over a set of wireless connected devices, random channel fluctuations, and the incumbent services simultaneously running on the same network affect the performance of distributed learning. In this paper, we investigate the interplay between distributed AI workflow and ultra-reliable low latency communication (URLLC) services running concurrently over a network. Using 3GPP compliant simulations in a factory automation use case, we show the impact of various distributed AI settings (e.g., model size and the number of participating devices) on the convergence time of distributed AI and the application layer performance of URLLC. Unless we leverage the existing 5G-NR quality of service handling mechanisms to separate the traffic from the two services, our simulation results show that the impact of distributed AI on the availability of the URLLC devices is significant. Moreover, with proper setting of distributed AI (e.g., proper user selection), we can substantially reduce network resource utilization, leading to lower latency for distributed AI and higher availability for the URLLC users. Our results provide important insights for future 6G and AI standardization.
Comments: Accepted in 2022 IEEE Global Communications Conference (GLOBECOM)
Subjects: Networking and Internet Architecture (cs.NI); Signal Processing (eess.SP)
Cite as: arXiv:2208.01352 [cs.NI]
  (or arXiv:2208.01352v1 [cs.NI] for this version)

Submission history

From: Milad Ganjalizadeh [view email]
[v1] Tue, 2 Aug 2022 10:46:50 GMT (508kb)

Link back to: arXiv, form interface, contact.