FedChain-BiGRU: An AI-driven Federated Deep Learning Framework for IoMT Intrusion Detection

Internet of Medical Things (IoMT) setups have changed remote patient tracking and digital diagnostics.Yet, this high connectivity exposes healthcare networks to hostile intrusion vectors.Centralized intrusion detection models can stop these threats, but centralized approaches require collecting traffic records at a central server, this paper reduces that requirement through local training.The model introduces a decentralized security framework that combines local training at client nodes with a tamper-evident validation ledger.First, we compress high-dimensional telemetry by 50% using an unsupervised Symmetric Deep Stacked Autoencoder.This handles the data noise and edge bandwidth limits.Next, these tightly packed latent features feed into a stacked, two-tier Bidirectional Gated Recurrent Unit (BiGRU) loop to capture complex temporal attack steps.Local edge nodes compute model weights independently, after which are combined through data-weighted Federated Averaging.The SHA-256 hashes of the aggregated weights are recorded using a lightweight metadata-only blockchain in order to capture unauthorized modification of aggregated model parameters during federated broadcasting.We validated (FedChain-BiGRU) architecture using the ToN-IoT dataset split across K= {10,15, 20} simulated edge clients.The system reached global accuracies of 99.14%, 99.17%, and 98.91% with corresponding F1-scores of 99.45%, 99.47%, and 99.30% under the IID setting for K= {10, 15, 20} respectively.Under the Non-IID setting, the best global accuracy for each client scale were obtained at α = 0.1, achieving global accuracies of 99.70%, 99.83%, 99.52% and corresponding F1-scores of 99.81%, 99.89%, and 99.69% for K= {10, 15, 20} respectively.The lightweight tracking blockchain maintained complete validity throughout.This proves the system provides highly accurate anomaly classification while keeping raw traffic records at local client nodes.

Authors

Institutions

Publication Details

Journal
International journal of intelligent engineering and systems
Published
2026-11-17
DOI
https://doi.org/10.22266/ijies2026.1231.67
Primary Topic
Network Security and Intrusion Detection
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

FedChain-BiGRU: An AI-driven Federated Deep Learning Framework for IoMT Intrusion Detection

Hasanain Ali Al Essa, Tahseen Ali Wotaifi, Amina Atiya Dawood, Balasem Allawi Hussain
2 links
International journal of intelligent engineering and systems
Network Security and Intrusion Detection
article

FedChain-BiGRU: An AI-driven Federated Deep Learning Framework for IoMT Intrusion Detection

Hasanain Ali Al Essa, Tahseen Ali Wotaifi, Amina Atiya Dawood, Balasem Allawi Hussain
article en

Abstract

Internet of Medical Things (IoMT) setups have changed remote patient tracking and digital diagnostics.Yet, this high connectivity exposes healthcare networks to hostile intrusion vectors.Centralized intrusion detection models can stop these threats, but centralized approaches require collecting traffic records at a central server, this paper reduces that requirement through local training.The model introduces a decentralized security framework that combines local training at client nodes with a tamper-evident validation ledger.First, we compress high-dimensional telemetry by 50% using an unsupervised Symmetric Deep Stacked Autoencoder.This handles the data noise and edge bandwidth limits.Next, these tightly packed latent features feed into a stacked, two-tier Bidirectional Gated Recurrent Unit (BiGRU) loop to capture complex temporal attack steps.Local edge nodes compute model weights independently, after which are combined through data-weighted Federated Averaging.The SHA-256 hashes of the aggregated weights are recorded using a lightweight metadata-only blockchain in order to capture unauthorized modification of aggregated model parameters during federated broadcasting.We validated (FedChain-BiGRU) architecture using the ToN-IoT dataset split across K= {10,15, 20} simulated edge clients.The system reached global accuracies of 99.14%, 99.17%, and 98.91% with corresponding F1-scores of 99.45%, 99.47%, and 99.30% under the IID setting for K= {10, 15, 20} respectively.Under the Non-IID setting, the best global accuracy for each client scale were obtained at α = 0.1, achieving global accuracies of 99.70%, 99.83%, 99.52% and corresponding F1-scores of 99.81%, 99.89%, and 99.69% for K= {10, 15, 20} respectively.The lightweight tracking blockchain maintained complete validity throughout.This proves the system provides highly accurate anomaly classification while keeping raw traffic records at local client nodes.

International journal of intelligent engineering and systemsVol. 19(12)
University of Babylon (IQ)
Network Security and Intrusion Detection
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon