LIDIT: Low-Latency Intrusion Detection in IoMT Devices using TinyML
Bibliographic record
Abstract
The Internet of Medical Things (IoMT) is reshaping healthcare by facilitating real-time monitoring, diagnosis, and treatment through interconnected devices and systems. However, the proliferation of resource-constrained IoMT devices introduces substantial cybersecurity challenges. Due to limited computational and energy resources, conventional security mechanisms such as complex encryption algorithms and robust firewalls are often infeasible. This poses serious risks in critical healthcare applications where delayed threat detection can lead to life-threatening outcomes. To address these pressing challenges, this research presents LIDIT, a novel anomaly-based intrusion detection framework with specialized feature segmentation designed for resource-constrained environments using TinyML. Our approach employs a multi-branch LSTM-autoencoder model trained exclusively on benign traffic, utilizing an input segmentation strategy based on session-level, TCP flags, and time-window features to capture fine-grained temporal as well as contextual patterns in network behavior. We evaluated the model on the CICIoMT2024 and IoMT-TrafficData dataset and demonstrated that our proposed segmentation framework improves anomaly detection performance over unified models. The best-performing model achieved an accuracy of 0.9990 and an F1-score of 0.9988 with a recall of 0.9995 for the CICIoMT2024 dataset. Post-training quantization using FLOAT16 and INT8 further significantly reduced the model sizes, making it suitable for real-time deployment. The system was successfully deployed on a Raspberry Pi Zero 2 W and tested under a live SYN flood attack, detecting anomalies in real time with an average inference time of 10.25 milliseconds. These results confirm the effectiveness, efficiency, and deployability of LIDIT as a lightweight, low-latency intrusion detection solution for modern healthcare IoT systems.
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 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 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".