TSM-NIDS: A time-series mixer-based intrusion detection system for IoT networks

Area of Science:

• Cybersecurity
• Network Security
• Internet of Things (IoT)

Background:

• The proliferation of IoT devices creates significant cybersecurity challenges.
• Traditional Intrusion Detection Systems (IDS) struggle with temporal dependencies in IoT network traffic.
• Existing methods often fail to capture crucial sequential patterns for effective threat detection.

Purpose of the Study:

• To propose TSM-NIDS, an adaptation of the TSMixer architecture for anomaly detection in IoT networks.
• To address the limitations of traditional IDS in analyzing time-series IoT data.
• To enhance the detection of cyber threats in diverse IoT environments.

Main Methods:

• An All-MLP (Multi-Layer Perceptron) design is employed for both time mixing and feature mixing.
• The TSMixer architecture is adapted for cybersecurity anomaly detection.
• Evaluation is performed using the publicly available TON-IoT dataset.

Main Results:

• TSM-NIDS demonstrates superior performance compared to existing state-of-the-art approaches.
• The system effectively learns sequential patterns and cross-feature dependencies in network traffic.
• Significant improvements in anomaly detection accuracy for IoT networks are achieved.

Conclusions:

• TSM-NIDS offers a promising solution for enhancing the security of IoT networks.
• The proposed method effectively captures temporal dependencies for robust intrusion detection.
• This research highlights the potential of TSMixer adaptations in cybersecurity applications.