Related Experiment Video
Updated: Jun 28, 2025
06:37
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
Published on: December 15, 2023
3.7K
OOA-modified Bi-LSTM network: An effective intrusion detection framework for IoT systems.
Siva Surya Narayana Chintapalli1, Satya Prakash Singh1, Jaroslav Frnda2,3
1Department of Computer Science and Engineering, Birla Institute of Technology, Mesra, Ranchi, India.
Heliyon
|April 22, 2024
Summary
This study introduces an enhanced Intrusion Detection System (IDS) for the Internet of Things (IoT) using Osprey Optimization Algorithm (OOA) for feature selection and a modified Bi-directional Long Short Term Memory (Bi-LSTM) for attack classification, achieving high accuracy.
Area of Science:
- Cybersecurity
- Computer Networks
- Artificial Intelligence
Background:
- Internet of Things (IoT) systems generate vast traffic data, increasing vulnerability to cyber intrusions.
- Existing Intrusion Detection Systems (IDS) face challenges with irrelevant features and slow training.
- Efficient IDS are crucial for ensuring the reliability, integrity, and security of IoT environments.
Purpose of the Study:
- To develop an enhanced Intrusion Detection System (IDS) for Internet of Things (IoT) security.
- To address challenges in intrusion detection, including feature relevance and training speed.
- To improve the accuracy and efficiency of identifying and classifying network intrusions.
Main Methods:
- Implemented Osprey Optimization Algorithm (OOA) for effective feature selection from network traffic data.
- Utilized a modified Bi-directional Long Short Term Memory (Bi-LSTM) network with Exponential Linear Unit (ELU) activation function for intrusion classification.
- Evaluated the proposed framework on N-BaIoT, CICIDS-2017, and ToN-IoT datasets.
Main Results:
- Achieved high detection accuracies: 99.98% on N-BaIoT, 99.97% on CICIDS-2017, and 99.88% on ToN-IoT.
- Demonstrated superior performance compared to existing frameworks in terms of accuracy and interpretability.
- Showcased reduced processing time due to optimized feature selection and faster learning with ELU activation.
Conclusions:
- The proposed OOA-based feature selection and modified Bi-LSTM network significantly enhance IoT Intrusion Detection System performance.
- The framework offers a robust solution for detecting and classifying various intrusion attacks with high precision.
- This approach provides a more interpretable and computationally efficient method for securing IoT systems.
