An efficient intrusion detection model based on convolutional spiking neural network.

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Computer Engineering

Background:

• Intrusion detection systems (IDS) are crucial for system integrity.
• Internet of Things (IoT) devices present unique challenges due to resource constraints.
• Deep learning (DL) and spiking neural networks (SNNs) show promise for IDS but often lack efficiency in constrained environments.

Purpose of the Study:

• To propose a lightweight and effective intrusion detection model for resource-constrained environments.
• To address the inefficiency of current SNN-based solutions in low-power, low-computation scenarios.
• To develop a model that balances resource usage with high classification accuracy.

Main Methods:

• Integration of Spiking Neural Networks (SNNs) and Convolutional Neural Networks (CNNs) through rational algorithm design.
• Development of a lightweight architecture optimized for minimal resource consumption.
• Evaluation against state-of-the-art models using comprehensive performance metrics.

Main Results:

• The proposed model significantly reduces resource usage compared to existing methods.
• High classification accuracy is maintained despite the lightweight design.
• Demonstrated superior adaptability in environments with limited computational and energy resources.

Conclusions:

• The developed SNN-CNN hybrid model offers an efficient solution for intrusion detection in IoT.
• The model effectively addresses the performance limitations of SNNs in resource-constrained settings.
• This approach provides a viable pathway for enhancing IoT security without compromising device capabilities.