Jamshed ALi Shaikh , Chengliang Wang , Saifullah , Muhammad Wajeeh Us Sima , Muhammad Arshad , Waheed Ul Asar Rathore
{"title":"基于内存反馈变压器的IoMT医疗网络入侵检测系统","authors":"Jamshed ALi Shaikh , Chengliang Wang , Saifullah , Muhammad Wajeeh Us Sima , Muhammad Arshad , Waheed Ul Asar Rathore","doi":"10.1016/j.iot.2025.101597","DOIUrl":null,"url":null,"abstract":"<div><div>Transformers, while effective at capturing spatial relationships through self-attention mechanisms, typically rely on LSTM networks only at the end to model sequential dependencies. This limits their ability to fully exploit temporal relationships across all layers. Such limitations impact the performance of Intrusion Detection Systems (IDS) in Internet of Medical Things (IoMT) environments , where accurate analysis of patient data is essential for detecting known attack signatures, zero-day anomalies, monitoring health trends, and securing healthcare networks. To address these challenges, we propose the Memory Feedback Transformer (MF-Transformer), which integrates Memory Feedback LSTM (MF-LSTM) throughout the entire Transformer architecture to capture and propagate temporal dependencies at every layer. The MF-Transformer first computes spatial-to-spatial relationships by analyzing correlations between features within the same time step, then incorporates spatial-to-temporal relationships by integrating the hidden state from the MF-LSTM to capture temporal dynamics via a feedback loop. By combining spatial and temporal patterns, the MF-Transformer retains long-term dependencies, tracks temporal dynamics effectively, and enhances anomaly detection, identifying both short-term deviations and long-term trends. Comprehensive evaluations on three publicly available datasets, WUSTL-EHMS-2020, ECU-IoHT, and X-IIoTID demonstrate the superior performance of the proposed MF-Transformer, achieving accuracy rates of 99.88%, 99.42%, and 99.12% for signature detection, and 99.98%, 99.71%, and 99.18% for anomaly detection, respectively.</div></div>","PeriodicalId":29968,"journal":{"name":"Internet of Things","volume":"32 ","pages":"Article 101597"},"PeriodicalIF":6.0000,"publicationDate":"2025-05-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Memory feedback transformer based intrusion detection system for IoMT healthcare networks\",\"authors\":\"Jamshed ALi Shaikh , Chengliang Wang , Saifullah , Muhammad Wajeeh Us Sima , Muhammad Arshad , Waheed Ul Asar Rathore\",\"doi\":\"10.1016/j.iot.2025.101597\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Transformers, while effective at capturing spatial relationships through self-attention mechanisms, typically rely on LSTM networks only at the end to model sequential dependencies. This limits their ability to fully exploit temporal relationships across all layers. Such limitations impact the performance of Intrusion Detection Systems (IDS) in Internet of Medical Things (IoMT) environments , where accurate analysis of patient data is essential for detecting known attack signatures, zero-day anomalies, monitoring health trends, and securing healthcare networks. To address these challenges, we propose the Memory Feedback Transformer (MF-Transformer), which integrates Memory Feedback LSTM (MF-LSTM) throughout the entire Transformer architecture to capture and propagate temporal dependencies at every layer. The MF-Transformer first computes spatial-to-spatial relationships by analyzing correlations between features within the same time step, then incorporates spatial-to-temporal relationships by integrating the hidden state from the MF-LSTM to capture temporal dynamics via a feedback loop. By combining spatial and temporal patterns, the MF-Transformer retains long-term dependencies, tracks temporal dynamics effectively, and enhances anomaly detection, identifying both short-term deviations and long-term trends. Comprehensive evaluations on three publicly available datasets, WUSTL-EHMS-2020, ECU-IoHT, and X-IIoTID demonstrate the superior performance of the proposed MF-Transformer, achieving accuracy rates of 99.88%, 99.42%, and 99.12% for signature detection, and 99.98%, 99.71%, and 99.18% for anomaly detection, respectively.</div></div>\",\"PeriodicalId\":29968,\"journal\":{\"name\":\"Internet of Things\",\"volume\":\"32 \",\"pages\":\"Article 101597\"},\"PeriodicalIF\":6.0000,\"publicationDate\":\"2025-05-03\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Internet of Things\",\"FirstCategoryId\":\"94\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S2542660525001106\",\"RegionNum\":3,\"RegionCategory\":\"计算机科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"COMPUTER SCIENCE, INFORMATION SYSTEMS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Internet of Things","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2542660525001106","RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, INFORMATION SYSTEMS","Score":null,"Total":0}
Memory feedback transformer based intrusion detection system for IoMT healthcare networks
Transformers, while effective at capturing spatial relationships through self-attention mechanisms, typically rely on LSTM networks only at the end to model sequential dependencies. This limits their ability to fully exploit temporal relationships across all layers. Such limitations impact the performance of Intrusion Detection Systems (IDS) in Internet of Medical Things (IoMT) environments , where accurate analysis of patient data is essential for detecting known attack signatures, zero-day anomalies, monitoring health trends, and securing healthcare networks. To address these challenges, we propose the Memory Feedback Transformer (MF-Transformer), which integrates Memory Feedback LSTM (MF-LSTM) throughout the entire Transformer architecture to capture and propagate temporal dependencies at every layer. The MF-Transformer first computes spatial-to-spatial relationships by analyzing correlations between features within the same time step, then incorporates spatial-to-temporal relationships by integrating the hidden state from the MF-LSTM to capture temporal dynamics via a feedback loop. By combining spatial and temporal patterns, the MF-Transformer retains long-term dependencies, tracks temporal dynamics effectively, and enhances anomaly detection, identifying both short-term deviations and long-term trends. Comprehensive evaluations on three publicly available datasets, WUSTL-EHMS-2020, ECU-IoHT, and X-IIoTID demonstrate the superior performance of the proposed MF-Transformer, achieving accuracy rates of 99.88%, 99.42%, and 99.12% for signature detection, and 99.98%, 99.71%, and 99.18% for anomaly detection, respectively.
期刊介绍:
Internet of Things; Engineering Cyber Physical Human Systems is a comprehensive journal encouraging cross collaboration between researchers, engineers and practitioners in the field of IoT & Cyber Physical Human Systems. The journal offers a unique platform to exchange scientific information on the entire breadth of technology, science, and societal applications of the IoT.
The journal will place a high priority on timely publication, and provide a home for high quality.
Furthermore, IOT is interested in publishing topical Special Issues on any aspect of IOT.