AI-Driven Biometric Authentication in IoT Systems: A Survey on Trade-Offs, Modalities, and Comparative Performance With Non-AI Approaches
- 14 October 2025
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. CFP25TZ1-A, 1-8
Abstract
Biometric authentication has emerged as a critical security mechanism for IoT ecosystems, balancing usability and robustness against evolving threats. This survey comprehensively analyzes 24 peer-reviewed papers to evaluate the role of artificial intelligence (AI) in enhancing biometric authentication for IoT systems, contrasting its performance with traditional non-AI approaches. We categorize frameworks by biometric modality (fingerprint, facial recognition, ECG), IoT layer (edge, application), and security mechanisms (cryptography, multifactor authentication), with a focus on accuracy, latency, and cost tradeoffs. Key findings reveal that 42 % of systems leverage AI, achieving superior accuracy (for example 99.99% for facial recognition in AI-driven systems vs. 93.05 % in non-AI) but often incurring higher latency (for example 12s for AI-based HD-sEMG vs. $\mathbf{1. 5 0 2 ~ m s}$ for non-AI Rabin cryptosystems). Non-AI systems dominate in low-resource environments (for example edge-layer fingerprint recognition) through lightweight cryptography (AES, Shamir Secret Sharing) and fuzzy extractors, prioritizing cost efficiency and real-time performance. AI excels in dynamic threat detection (for example honey templates with automated alarms) and complex biometrics (ECG, gait), though computational overhead remains a barrier for resource-constrained devices. We identify critical gaps, including the lack of standardized benchmarks for accuracy/latency and limited adoption of continuous authentication. This survey uniquely focuses on the comparative analysis of AI and non-AI biometric systems across different IoT layers and resource constraints-an angle underexplored in prior surveys This survey provides a practical insight for researchers and developers to navigate trade-offs between AI-driven innovation and traditional security in IoT biometrics.
Keywords
This publication has 19 references indexed in Scilit:
- Biometry-based verification system with symmetric key generation method for internet of things environmentsScientific Reports, 2025
- A biometric and PUF-based authentication with Preserving anonymity in smart grid environmentAin Shams Engineering Journal, 2024
- An efficient authentication scheme syncretizing physical unclonable function and revocable biometrics in Industrial Internet of ThingsJournal of King Saud University Computer and Information Sciences, 2024
- ECG Biometric Authentication Using Self-Supervised Learning for IoT Edge SensorsIEEE Journal of Biomedical and Health Informatics, 2024
- Robust Rabin Cryptosystem Biometric Privacy Preserving Authentication Technique for IoT Enabled Devices Over CloudSN Computer Science, 2024
- A Survey on Security, Privacy, Trust, and Architectural Challenges in IoT SystemsIEEE Access, 2024
- Multimodal cancelable biometric authentication system based on EEG signal for IoT applicationsJournal of Optics, 2023
- Survey of Secure Communications of Internet of Things with Artificial IntelligenceIEEE Internet of Things Magazine, 2022
- HIAuth: A Hierarchical Implicit Authentication System for IoT Wearables Using Multiple BiometricsIEEE Access, 2021
- Robust multimodal biometric authentication on IoT device through ear shape and arm gestureMultimedia Tools and Applications, 2021