Jammer Detection in Autonomous Vehicles with Machine Learning - Master’s Thesis - Jammer attack classification using KNN, Random Forest, and XGBoost

Aug 3, 2026

Wireless communication systems of autonomous vehicles can be exposed to jammer attacks, leading to operational failures and security threats. This study classifies jammer attack scenarios using machine learning to determine appropriate countermeasures. A simulated dataset is constructed with key parameters including RSSI, SNR, PDR, and estimated relative speed. KNN, Random Forest, and XGBoost models are trained and evaluated comparatively for jammer detection performance. Results indicate that all three models achieve similar accuracy, consistently above 90%, demonstrating that machine learning can enable reliable identification of jammer attacks in autonomous vehicle communication.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages79

Reading Time3 h 19 min

AccessFree to download

List of Tables

List of Figures

Chapter 1 Introduction

Chapter 2 Literature Review

Chapter 3 Research Methodology

Chapter 4 Results

Chapter 5 Conclusion

References

Appendix A Confusion Matrices

Appendix B Weights & BiaSES

Middle East Technical University Informatics Institute JAMMER DETECTION IN AUTONOMOUS VEHICLES WITH MACHINE LEARNING Advisor Name: Prof. Dr. Tuğba TAŞKAYA TEMİZEL (METU) Student Name: Erdoğan Mert DEMİRYÜREK (IS) Orta Doğu Teknik Üniversitesi Enformatik Enstitüsü MAKİNE ÖĞRENİMİ İLE OTONOM ARAÇLARDA JAMMER TESPİTİ Danışman Adı: Prof. Dr. Tuğba TAŞKAYA TEMİZEL (ODTÜ) Öğrenci Adı: Erdoğan Mert DEMİRYÜREK (BS) | REPORT DOCUMENTATION PAGE | | | | --- | --- | --- | | 1. AGENCY USE ONLY (Internal Use) | 2. REPORT DATE
10.01.2025 | | | 3. TITLE AND SUBTITLE
JAMMER DETECTION IN AUTONOMOUS VEHICLES WITH MACHINE LEARNING | | | | 4. AUTHOR (S)
Erdoğan Mert DEMİRYÜREK | 5. REPORT NUMBER (Internal Use) | | | 6. SPONSORING/ MONITORING AGENCY NAME(S) AND SIGNATURE(S) | | | | 7. SUPPLEMENTARY NOTES | | | | 8. ABSTRACT (MAXIMUM 200 WORDS)
Wireless communication systems of autonomous vehicles may be vulnerable to threats such as jammer attacks. As a result of these attacks, operational failures and security problems may occur. This study aims to classify which jammer attacking scenario occurs, using machine learning in order to select the necessary precaution to be taken against jammer attacks. A simulated dataset consisting of parameters such as RSSI, SNR, PDR and estimated relative speed was used in the study. KNN, Random Forest and XGBoost models were used for jammer detection and their performances were compared. The results showed that all of the models, KNN, Random Forest, and XGBoost models has similar accuracy results which is above 90% . The results show that accurate detection of jammer attacks can be achieved with the machine learning algorithms | | | | 9. SUBJECT TERMS | | 10. NUMBER OF PAGES
74 | TABLE OF CONTENTS LIST OF TABLES...................................................................................................................... ii LIST OF FIGURES................................................................................................................... iii CHAPTER 1 INTRODUCTION ............................................................................................... 1 CHAPTER 2 LITERATURE REVIEW ..................................................................................... 3 CHAPTER 3 RESEARCH METHODOLOGY......................................................................... 7 CHAPTER 4 RESULTS........................................................................................................... 28 CHAPTER 5 CONCLUSION .................................................................................................. 36 REFERENCES......................................................................................................................... 37 APPENDIX A CONFUSION MATRICES .............................................................................. 40 APPENDIX B WEIGHTS & BIASES..................................................................................... 51 LIST OF TABLES Table 1 Train and Test Set size for Dataset 1 ........................................................................... 19 Table 2 Train and Test Set size for Dataset 2 ........................................................................... 20 Table 3 Train Set size for Dataset 1 and Test Set size for Dataset 2 ........................................ 20 Table 4 Model Combinations ................................................................................................... 25 Table 5 Hyperparameters.......................................................................................................... 26 Table 6 Accuracy Results of Each Model ................................................................................ 29 Table 7 Performance Metrics ofKNN Train: 15 m/s Test 15 m/s, No VRS, No Normalization..........................................................................................

FAQs

What problem does the study address?

The study addresses how wireless communication in autonomous vehicles can be disrupted by jammer attacks and how to detect which jammer scenario occurs.

Which machine learning models are used for jammer detection?

KNN, Random Forest, and XGBoost are used, and their performances are compared across experiments.

What inputs and dataset characteristics does the research use?

A simulated dataset includes RSSI, SNR, PDR, and estimated relative speed to represent communication and motion-related conditions under different jammer scenarios.

Aug 3, 2026

Upload by River Wang