Automobile odometer fraud prevention with the implementation of blockchain and deep learning

Abstract

Automobile odometer fraud prevention has been an issue for the automobile industry for some time. The estimated annual financial damages from this fraud exceed one billion dollars. It is necessary to have a solution for vehicle data that is both secure and immutable. In response to the requirements, we have chosen to implement Blockchain technology to combat automotive odometer fraud. In our study, we demonstrate and describe a comprehensive fraud prevention solution based on Deep Learning & Ethereum Blockchain technology. Previously, there have been research that used blockchain to prevent odometer fraud. However, all of these systems had one flaw that may jeopardize the system's security even before integrating blockchain. The OBD2 port, which is utilized to obtain the odometer reading, necessitates the presence of a physical adapter in the vehicle. This raises major security concerns since any tampering with the adaptor would result in odometer data alteration even before the data is deployed on the blockchain. As a result, we propose a novel solution that addresses this issue. We used state-of-the art object detection models based on CNNs to extract the odometer reading from the image and cross-validate it with the odometer reading from the adapter. The odometer reading is then uploaded to the blockchain leveraging smart contracts. We developed a comprehensive system architecture to prevent odometer fraud and addressed security risks associated with OBD2 adapters used in the process of extracting odometer readings