Related Experiment Video
Updated: May 21, 2025
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
448
A high performance hybrid LSTM CNN secure architecture for IoT environments using deep learning
Priyanshu Sinha1, Dinesh Sahu2, Shiv Prakash3
1Department of Electronics and Communication, University of Allahabad, Prayag Raj, Uttar Pradesh, India.
Scientific Reports
|March 21, 2025
Summary
This study introduces an Advanced LSTM-CNN framework for real-time IoT intrusion detection. The model achieves high accuracy and robustness against cyber threats, improving IoT device security.
Area of Science:
- Cybersecurity
- Artificial Intelligence
- Internet of Things (IoT)
Background:
- The proliferation of Internet of Things (IoT) devices has introduced significant security vulnerabilities, necessitating advanced intrusion detection systems.
- Existing deep learning models face challenges in effectively identifying diverse cyber threats in real-time IoT environments.
Purpose of the Study:
- To propose and evaluate an Advanced Long Short-Term Memory (LSTM)-Convolutional Neural Network (CNN) Secure Framework for optimizing real-time intrusion detection in IoT networks.
- To enhance the identification of various cyber attack typologies, including Distributed Denial of Service (DDoS), botnet, reconnaissance, and data exfiltration.
Main Methods:
- Development of a hybrid LSTM-CNN model integrating LSTM layers for temporal dependency learning and CNN layers for spatial feature decomposition.
- Utilizing the BoT-IoT dataset, which encompasses a wide range of cyber attack scenarios.
- Performance evaluation against other deep learning models (CNN, RNN, Standard LSTM, BiLSTM, GRU) and assessment of robustness against adversarial attacks using SHAP for feature importance analysis.
Main Results:
- The proposed LSTM-CNN model achieved high performance metrics: 99.87% accuracy, 99.89% precision, and 99.85% recall, with a low false positive rate of 0.13%.
- The model demonstrated superior performance compared to conventional deep learning models.
- Achieved 90.2% accuracy under adversarial attack conditions, indicating significant robustness.
- Feature importance analysis identified packet size, connection duration, and protocol type as key indicators for threat detection.
Conclusions:
- The Advanced LSTM-CNN Secure Framework offers a robust and efficient solution for real-time intrusion detection in IoT environments.
- The model's high accuracy, precision, and recall, coupled with its resilience to adversarial attacks, make it suitable for practical security applications.
- The findings suggest that this hybrid approach can significantly enhance IoT device security by providing reliable threat detection with minimal false alarms.
