A new method for the detection and identification of the replay attack on cars using SDR technology and classification algorithms

Abstract

Context: Currently, a wide variety of cars use Remote Keyless Entry (RKE) to activate the security system by radio frequency. Problem: RKE systems have security vulnerabilities with attacks such as brute force, replay, spoofing, and replay. Objective: This study proposes a replay-type attack detection system using SDR and intelligent algorithms. Method: The present dataset was obtained from an RKES type Key-fob of a 2016 model vehicle taken with different antenna polarizations at different distances. There were used three test and configuration systems based on two neural network identification models. Results: The proposed designs with the different classification models were able to detect replay-type vulnerabilities. In the first system, retransmitted indicators were detected with a maximum identification percentage of 90.9% and original signs of 90.9%; for the second system, 100% of the retransmitted signals and 89% of the original ones were detected, and for the third system, detected 100% of the retransmitted signals and 89.09% of the first signs. Conclusion: The models used in the system's design showed an identification percentage of more than 87% in all cases, managing to detect and identify the replay-type attack on the vehicles. System 1 presented better results with the VGG 16 system than in system 2, within the VGG 16 and SVM models having identical identification percentages obtained and, in system 3, which has the characteristics of the previous two, attained better results with SVM. The reception of signals with SDR is an effective tool for detecting vulnerabilities in vehicle security systems.

Keywords

• Replay attack
• Software defined radio
• VGG
• SVM
• Attack