Title: Increasing network security using an enhanced hybrid deep intrusion detection model

Authors: Jiacheng Wu; Tingting Jiang; Juan Li; Wujun Mei

Addresses: Taizhou Vocational College of Science and Technology, No. 288, Jiamu Rd., Taizhou, Zhejiang, China ' Taizhou Vocational College of Science and Technology, No. 288, Jiamu Rd., Taizhou, Zhejiang, China ' Taizhou Vocational College of Science and Technology, No. 288, Jiamu Rd., Taizhou, Zhejiang, China ' Taizhou Institute of Zhejiang University, No. 618, West Section of Shifu Avenue, Taizhou, Zhejiang, China

Abstract: Internet of things (IoT) systems have recently seen a widespread use of machine learning (ML) methodologies for intrusion detection systems (IDSs). LSTM and GRU models, which are the RNNs, are used to identify the many kinds of threats that may occur in IoT systems. The Harris hawk optimisation and fractional derivative mutation methods are used in this study to perform feature choices. To evaluate the suggested technique, datasets that are accessible to the public were used. The empirical analysis revealed that the proposed method is superior to the other related approaches in accuracy and efficiency. The proposed model makes use of several databases. The proposed model attained a maximum accuracy of 100% in identifying attacks such as denial of service, exploits, generic, reconnaissance, and shellcode attacks respectively. This model provides a 99.7% accuracy rate.

Keywords: intrusion detection systems; IDSs; machine learning; metaheuristics deep learning; long short-term memory.

DOI: 10.1504/IJESDF.2026.153332

International Journal of Electronic Security and Digital Forensics, 2026 Vol.18 No.3, pp.323 - 335

Received: 11 Jan 2024
Accepted: 19 Mar 2024
Published online: 01 May 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article