This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/124503
Description
Title
Federated learning in drone-based systems
Author(s)
Piramuthu, Otto Benjamin
Issue Date
2024-04-08
Director of Research (if dissertation) or Advisor (if thesis)
Caesar, Matthew C
Department of Study
Computer Science
Discipline
Computer Science
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
M.S.
Degree Level
Thesis
Keyword(s)
Drone
Federated Learning
Abstract
Drones have unique characteristics that allow them to perform tasks that otherwise cannot be accomplished with such efficiency in a cost-effective manner. In many applications, drones often are part of a system with several other drones that help perform related operations. In such a setup, it is beneficial for the drones to learn from the experience of one another. However, sharing sensitive data could be an issue when these drones capture data that belong to different entities (e.g., neighboring farms with different owners who are hesitant to share fine granular data). Federated learning is a natural choice in applications where drones do not want to share their data with any other entity. The federated learning framework comprises several clients (drones) and a server (a base station), where each drone generates a local model with its data and then shares the local model parameter updates with the server, which aggregates these to generate the global model. Operational and strategic constraints such as communication between drones and base station while the drones are mobile, limited and slow communication channels, uncooperative drones, and information aggregation from multiple drones as well as associated scalability challenges need to be addressed for smooth operation of drone-based systems. We study three elements in this setup. Specifically, we study the amount of shared information (scalability), when to share, and how to aggregate such shared information. Each of these are significant elements that need to be carefully considered for seamless incorporation of federated learning in drone-based systems. Our results indicate that the number of drones and the amount of information they share with the server (base station) are complements where an increase in one compensates for a decrease in the other. We derive bounds for the conditions under which drones want to share their local model updates with the server. For privacy reasons, when drones decide to share only range values instead of exact values of their local model parameters, we observe the decisions based on global model outputs to be different.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
