Biometric Authentication Using sPUF and IPFS for Electric Vehicles
B. Satya Swaroop1, Anshika Jha2, T. Jhansi3, M. Sai Sandeep4, V. Hima Varshini5, D. Nirmal6
1Assistant Professor, Computer Science and Engineering Department, Raghu Engineering College, Visakhapatnam, India.
2,3,4,5,6 Computer Science and Engineering Department, Raghu Institute of Technology, Visakhapatnam, India.
Abstract
This project presents a Smart Biometric Authentication System for Electric Vehicles, designed to provide a software-only, multi-factor biometric authentication solution that combines face recognition and fingerprint verification with advanced cryptographic protocols. The system leverages simulated Physical Unclonable Functions (sPUF) for hardware-bound challenge-response pair generation, fuzzy extractors with Reed-Solomon error-correcting codes for noise-tolerant biometric key derivation, and decentralized storage through IPFS via Pinata for tamper-resistant enrollment data management.
The proposed system implements a complete authentication pipeline consisting of biometric capture using WebRTC and MediaPipe for facial landmark extraction (478 3D landmarks) and OpenCV ORB for fingerprint descriptor extraction (500 keypoint descriptors), followed by cryptographic processing through the sPUF and fuzzy extractor modules. All enrollment data is stored on IPFS as content-addressed JSON documents, while every authentication event, EV action, and access revocation is immutably logged on a blockchain via smart contracts deployed on Hardhat Network.
The system is built as a full-stack web application with a Python FastAPI backend for biometric and cryptographic processing, a Next.js 16 frontend with an animated pipeline visualization dashboard, and Solidity smart contracts for on-chain verification. The entire application is deployable as a single project on Vercel using the experimental Services feature.
Experimental results demonstrate that the combined biometric system achieves a 96% overall authentication accuracy with a False Accept Rate (FAR) of 1% and False Reject Rate (FRR) of 3%, significantly outperforming single-modality approaches. The blockchain-based audit trail ensures complete transparency and non-repudiation of all vehicle access events.
Keywords
Biometric Authentication, Electric Vehicles, Physical Unclonable Functions, Fuzzy Extractor, IPFS, Blockchain, Smart Contracts, Face Recognition, Fingerprint Verification, Reed-Solomon Codes, Decentralized Storage.
