Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.

Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.

Author ORCID Identifier

https://orcid.org/0000-0002-8509-9606

AccessType

Open Access Dissertation

Document Type

dissertation

Degree Name

Doctor of Philosophy (PhD)

Degree Program

Electrical and Computer Engineering

Year Degree Awarded

2023

Month Degree Awarded

May

Subject Categories

Systems and Communications

Abstract

The development of processors and sensor technologies and the advent of artificial intelligence (AI) have led to the growth of more capable and affordable mobile intelligent agents like unmanned aerial vehicles (UAVs) which have numerous applications in areas such as construction, search and rescue operations, oil and gas, agriculture, etc. The use of AI and machine learning (ML) algorithms in UAVs has enabled real-time data-driven decision-making frameworks but poses numerous challenges including high-impact decision-making, security, energy management, communication, and automation, to name a few. In this dissertation, we propose solutions to some of these problems. First, we give our attention to the probabilistic decision-making problem with an emphasis on the problems where the law of large numbers (LLN) does not apply. These problems are referred to as ``high-impact" problems. We present a comprehensive decision-making framework based on change-of-probability measures that addresses both the high-impact and ordinary (non-high-impact) problems, offering a systematic approach for rational decision-making. In other words, the proposed framework addresses the lack of concrete frameworks for rational decision-making, especially in the high-impact scenario. Subsequently, we move on to discuss the privacy issue, with particular attention paid to location privacy, the lack of which presents significant risks for users. In this regard, we first propose a model-free privacy-preserving mechanism while imposing a continuity constraint on the location data to avoid leakages from abrupt jumps. We introduce an algorithm where the privacy and utility guarantees as well as the continuity constraint are met. The location privacy problem is then investigated in UAV applications, in particular, UAV package delivery and the Internet of Things (IoT) data collection. The goal is to protect the location privacy of a UAV destination's location privacy from an adversary observing its trajectory. Considering performance metrics corresponding to each application, we analyze the privacy-utility tradeoffs. In particular, we provide numerical results for the tradeoff between privacy and mean peak age of information (PAoI) in the IoT data collection application.

DOI

https://doi.org/10.7275/35650349

Download

Share

COinS