AI-Driven Device Fingerprinting Using On-Chip Monitoring Sensors: A Novel Time Series-Based Approach
Bibliographic record
Abstract
The surge in current technological trends is leading to a rapid daily increase in the number of electronic devices. Within the broad spectrum of interconnected ecosystems, such as the Internet of Things (IoT) and cyber-physical systems (CPSs), authentication services play a pivotal role in ensuring trust and security. In line with this, hardware identification of devices is increasingly becoming an integral part of security frameworks, whether for protection against adversarial attacks or as an anticounterfeiting measure for integrated circuit (IC) verification. For this purpose, device fingerprinting (DFP) based on intrinsic physical variations of hardware has proven to be a highly reliable asset. In this article, the ubiquitous presence of on-chip sensors for internal monitoring has remained largely unexplored until now. In this work, we present a direct lightweight approach to the exploitation of these sensors across various scenarios. Through an artificial intelligence (AI)-driven methodology, we explore the use of diverse models [XGBoost, attentional convolutional neural network (CNN), and bidirectional long short-term memory (Bi-LSTM)] for device identification using fixed-length temperature-voltage time series pairs, evaluating the impact of different lengths on the identification process. The experimental results demonstrate unprecedented performance, achieving nearly 100% across all metrics obtained in the proposed scenarios. To assess the robustness of the solution, we utilized diverse datasets generated from the stimulation of electronic activity through workloads applied to 20 ultralow-power STM32L-DISCOVERY batteryless devices. Finally, we demonstrate the resilience of the solution under extreme conditions with five devices, subjected to undervolting, high- and low-temperature environments, and an accelerated aging test, reaffirming the previously obtained results.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".