All ETDs from UAB

Advisory Committee Chair

Ragib Hasan

Advisory Committee Members

Purushotham Bangalore

Mohammad Asadul Hoque

S M Riazul Islam

Sidharth Kumar

Document Type

Dissertation

Date of Award

2022

Degree Name by School

Doctor of Philosophy (PhD) College of Arts and Sciences

Abstract

Recently, Autonomous vehicles (AVs) have achieved tremendous improvement and started moving out to the road from laboratories and constrained testing environments. AVs are expected to significantly improve the urban transportation system by reducing accidents and drivers’ burdens. However, this complex cyber-physical system incurs some unique security challenges. Addressing such security issues is essential because AVs can directly impact the safety of human life. Current autonomous driving systems do not emphasize security issues. Therefore, sophisticated attack methodologies continue to evolve to threaten the safety of AVs. Although some research works have been proposed to enhance AV security, the holistic approach towards AV security has not been addressed until now. In this dissertation, we explore different techniques for enhancing the security, safety, and trustworthiness of connected autonomous vehicles. We propose a holistic security framework based on modern open-source autonomous driving systems using lightweight cryptographic techniques and autonomous driving components. The proposed framework presents a scheme for analyzing safety impact due to sensor spoofing attacks using a high-definition (HD) map and different sensor suites for multiple vehicle types. The framework identifies a security vulnerability in the adaptive cruise control feature in an industry-grade advanced driving assistance systems software. It proposes a safeguard for detecting such incidents using an HD map and global positioning system (GPS). The framework presents a secure and trustworthy forensic investigation framework that enables evidence collection from an autonomous driving system, maintaining secure provenance of evidence and obtaining evidence relevant to an incident. The framework proposes a security benchmarking framework that quantifies the severity of an attack mechanism or the effectiveness of a defense mechanism based on several security and performance metrics. Finally, the framework also consists of a protection mechanism that prevents authenticated but malicious vehicles in connected vehicle environments from sharing wrong information intentionally or unintentionally by proposing a trust management scheme. iii We implement prototypes of the proposed security schemes included in the framework using two industry grade autonomous driving and advanced diving assistance systems software, which are Autoware and Openpilot. We evaluate the performance of these schemes regarding efficiency, communication and computation overheads, and safety consequences. We also perform security analysis of our proposed schemes to demonstrate that the proposed framework is secure against strong adversarial scenarios.

Share

COinS