Analysis and Optimization about Federated Learning in Uploading Zone 


Vol. 48,  No. 2, pp. 266-274, Feb.  2023
10.7840/kics.2023.48.2.266


PDF
  Abstract

In this paper, we propose an Uploading Zone where devices can participate in federated learning, given they are distributed randomly. Here, the uploading zone is divided into the two regions, inner and outer regions, according to the quantization levels considering the path loss depending on device location. Some devices located in inner region utilize a sufficient number of bits for quantization because they are close to Base Station (BS), but the devices located in outer region use a fewer bits for quantization because they are far from BS. Hence we optimize the local update under this heterogeneous quantization condition, so as to maximize the global learning performance. To this end, we formulate an optimization problem to determine the optimum size of uploading zone that leads to the maximum learning performance. This finding can serve as an appropriate framework for AI-native IoT Networks which consist of a huge number of IoT sensors with diverse private data.

  Statistics
Cumulative Counts from November, 2022
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.


  Related Articles
  Cite this article

[IEEE Style]

S. H. Jeon and D. I. Kim, "Analysis and Optimization about Federated Learning in Uploading Zone," The Journal of Korean Institute of Communications and Information Sciences, vol. 48, no. 2, pp. 266-274, 2023. DOI: 10.7840/kics.2023.48.2.266.

[ACM Style]

Seong Hoon Jeon and Dong In Kim. 2023. Analysis and Optimization about Federated Learning in Uploading Zone. The Journal of Korean Institute of Communications and Information Sciences, 48, 2, (2023), 266-274. DOI: 10.7840/kics.2023.48.2.266.

[KICS Style]

Seong Hoon Jeon and Dong In Kim, "Analysis and Optimization about Federated Learning in Uploading Zone," The Journal of Korean Institute of Communications and Information Sciences, vol. 48, no. 2, pp. 266-274, 2. 2023. (https://doi.org/10.7840/kics.2023.48.2.266)
Vol. 48, No. 2 Index

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880