Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


Abiral-Thesis.pdf (5.74 MB)

ETD Abstract Container

Abstract Header

A Machine Learning Approach for Securing Autonomous and Connected Vehicles

Acharya, Abiral
http://rave.ohiolink.edu/etdc/view?acc_num=toledo1628099781846354

Abstract Details

2021, Master of Science, University of Toledo, Engineering (Computer Science).
With the rise of road congestion in modern cities, there are a lot of road traffic accidents. According to various studies, human drivers are at fault for over 90% of the accidents that occur on the road. Over the years, there have been huge advancements in the field of self-driving cars, also known as Autonomous Vehicles (AV), where there is little to no human involvement during driving. The main challenge AVs face is in their autonomy with the non-autonomous entities on the roads like pedestrians, other vehicles, and infrastructure. For a smooth road network, the vehicles communication is of major importance and the development of Vehicular Adhoc Network (VANET) is rapidly increasing. It is of utmost importance to create an efficient and secure routing protocol to route the safety messages within the vehicles in the network. In addition to this, it is also necessary to identify and prevent malicious messages in the network from interfering with effective and efficient communication. This thesis presents a brief survey on potential failures and attacks on AVs and VANETs. It performs a simulation study of VANET routing protocols: Optimized Link State Routing Protocol (OLSR), Ad-hoc On-demand Distance Vector (AODV), and Destination-Sequenced Distance-Vector Routing (DSDV) on a real-world map under different scenarios: low, medium, and high-density network. The results show OLSR performing better in delivering the basic safety messages within the network, and AODV performing better in terms of average routing goodput. The MAC/PHY overhead in the network depended upon the network density and DSDV and AODV came quite close in lowering the overhead. In addition, the thesis proposes a machine learning approach for identifying and preventing blackhole attacks in VANET. Different supervised machine learning classifiers are compared based on accuracy, F1-score, and Receiver Operating Characteristic (ROC) Area Under the Curve (AUC) score. The experiments show that the best performing model, Gradient Boosting Classifier, achieved an ROC AUC score of 0.971, an F1-score of 0.961, and classification accuracy of 0.9834.
Jared Oluoch (Advisor)
Ezzatollah Salari (Committee Member)
Junghwan Kim (Committee Member)
129 p.
Computer Science
autonomous vehicles, VANET, routing protocols, machine learning, security

Recommended Citations

Citations

  • Acharya, A. (2021). A Machine Learning Approach for Securing Autonomous and Connected Vehicles [Master's thesis, University of Toledo]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=toledo1628099781846354

    APA Style (7th edition)

  • Acharya, Abiral. A Machine Learning Approach for Securing Autonomous and Connected Vehicles. 2021. University of Toledo, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=toledo1628099781846354.

    MLA Style (8th edition)

  • Acharya, Abiral. "A Machine Learning Approach for Securing Autonomous and Connected Vehicles." Master's thesis, University of Toledo, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=toledo1628099781846354

    Chicago Manual of Style (17th edition)

toledo1628099781846354
724
© , all rights reserved.
This open access ETD is published by University of Toledo and OhioLINK.