Date of Award

2019

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical and Computer Engineering

Committee Chair

Seong-Moo Yoo

Committee Member

B. Earl Wells

Committee Member

W. David Pan

Committee Member

Dongsheng Wu

Committee Member

Murat M. Tanik

Subject(s)

Vehicular ad hoc networks (Computer networks), Wireless communication systems., Routing protocols (Computer network protocols), Machine learning.

Abstract

Vehicular Ad Hoc Networks (VANET) have a great potential to become the next technological paradigm shift to enable Intelligent Transportation Systems (ITS). Over the years, significant research progress has been made in VANETs to develop standards, protocols and applications to improve driver/passenger safety, traffic efficiency, and general cooperative awareness. Broadcast communications are the main critical information dissemination method in VANETs. Such wireless broadcasts empower highly dynamic network nodes in VANETs to quickly share pertinent information (emergency accident alert, road conditions, etc.) with each other. The highly dynamic nature of moving nodes and geographical constraints make the VANET topology a complex environment to operate such broadcast protocols. To this day, no single universal VANET broadcast routing protocol has been created, but rather a wide range of various protocols that solve specific challenges have been proposed. One of the main challenges that those broadcast protocols aim to solve is the issue of emergency information dissemination in VANETs. This dissertation proposes a fail-safe misbehavior detection and mitigation framework for the broadcast routing protocols in VANETs. The proposed framework will include a novel unintentional broadcast protocol misbehavior technique using entropy and machine learning classification. The fail-safe framework’s architecture consists of three major parts: local threshold monitoring, flow sampling and entropy-based detection, and finally the global verification using machine learning classification. The first part is implemented and configured at each node, where threshold monitoring applies to physical, MAC and other relevant network layer parameters. The second part is triggered based on the monitoring results and involves selective flow sampling with entropy analysis and optional post-detection action configurations. The final part of the framework involves Road Side Unit (RSU) based flow aggregation and efficient binary classification using the compressed machine learning model. The classification result analysis and extensive simulation results demonstrate that the framework performs well in terms of speed, overhead, accuracy, precision, and recall during the broadcast protocol misbehavior in VANETs.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.