An attack detection method based on deep learning for internet of things.

Area of Science:

• Computer Science
• Cybersecurity
• Artificial Intelligence

Background:

• Internet of Things (IoT) devices face increasing cyber threats and exponential growth in malicious network traffic.
• Existing attack detection methods struggle with complex attacks, leading to high false positive rates and performance limitations due to feature redundancy and class imbalance in IoT datasets.

Purpose of the Study:

• To develop a robust deep learning-based attack detection method for Internet of Things (IoT) environments.
• To overcome limitations of current methods, including high false positive rates and challenges with feature redundancy and class imbalance.

Main Methods:

• Feature selection using a genetic algorithm.
• Addressing class imbalance with a cost-sensitive function.
• Employing a hybrid Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) network for spatiotemporal feature extraction.

Main Results:

• The proposed method demonstrated superior performance on two benchmark IoT datasets.
• Effective enhancement of IoT attack detection capabilities was achieved.

Conclusions:

• The integrated deep learning approach, combining genetic algorithms, cost-sensitive learning, and CNN-LSTM networks, significantly improves IoT attack detection.
• This method offers a promising solution for enhancing the security of Internet of Things devices against sophisticated network attacks.