Network Attacker Detection Using a Hybrid Graph Neural Network and Temporal Convolutional Network Framework

In this work, we proposed a novel hybrid deep learning approach combining Graph Neural Networks (GNNs) and Temporal Convolutional Networks (TCNs) for effective detection of malicious activities in network environments based on system call sequences. By leveraging the spatial dependencies captured by GNNs and the temporal patterns modelled by TCNs, the approach robustly detects sophisticated and evolving cyberattacks that may not be easily identified using traditional methods. The dataset of benign and malicious system call traces provided a comprehensive basis for training and evaluation, and the data augmentation applied to the benign dataset improved generalization and addressed class imbalance. Experimental results show that the GNN-TCN hybrid model significantly outperforms conventional machine learning algorithms and other state-of-the-art deep learning models in accuracy, detection speed and adaptability to novel attack patterns. The ability to handle both spatial and temporal features positions the proposed system as a promising solution for next-generation network intrusion detection systems. In future work, we aim to expand the framework to more complex attack scenarios and larger datasets, explore real-time intrusion detection in operational environments, and integrate more advanced graph-based and temporal analysis techniques to further enhance robustness and efficiency.

[1] R. Janapati, C. Balaswamy and K. Soundararajan, “Enhancement of localized routing using CDPSO in WSN,” 2018 Conference on Signal Processing and Communication Engineering Systems (SPACES), Vijayawada, 2018, pp. 16-19. [2] I. Daanoune, A. Baghdad and A. Balllouk, “A comparative study between ACO-based protocols and PSO-based protocols in WSN,” 2019 7th Mediterranean Congress of Telecommunications (CMT), Fès, Morocco, 2019, pp. 1-4. [3] M. Shinde and D. C. Mehetre, “Black Hole and Selective Forwarding Attack Detection and Prevention in WSN,” 2017 International Conference on Computing, Communication, Control and Automation (ICCUBEA), Pune, 2017, pp. 1-6. [4] A. Karthikeyann, V. P. Arunachalam, S. Karthik and P. Dhivya, “Energy Efficient Structure Free and Location Based Routing Protocol in WSN,” 2018 International Conference on Soft-computing and Network Security (ICSNS), Coimbatore, 2018, pp. 1-5. [5] A. Mathur and T. Newe, “Medical WSN: Power, routing and selective forwarding defense,” 2015 13th International Conference on Telecommunications (ConTEL), Graz, 2015, pp. 1-6. [6] M. Saikia, U. K. Das and M. A. Hussain, “Secure energy aware multi-path routing with key management in wireless sensor network,” 2017 4th International Conference on Signal Processing and Integrated Networks (SPIN), Noida, 2017, pp. 310-315. [7] I. Ouafaa, E. Mustapha, K. Salah-ddine, L. Jalal and E. H. Said, “An advanced analysis on secure hierarchical routing protocols in wireless sensor network,” 2016 International Conference on Engineering & MIS (ICEMIS), Agadir, 2016, pp. 1-6. [8] A. V. Mathurkar and P. D. Dorge, “Performance Evaluation of Network Using AOMDV Routing Protocol,” 2020 8th International Conference on Reliability, Infocom Technologies and Optimization (ICRITO), Noida, India, 2020, pp. 479-483. [9] G. Kalnoor and J. Agarkhed, “QoS based multipath routing for intrusion detection of sinkhole attack in wireless sensor networks,” 2016 International Conference on Circuit, Power and Computing Technologies (ICCPCT), Nagercoil, 2016, pp. 1-6. [10] A. Patle and N. Gupta, “Vulnerabilities, attack effect and different security scheme in WSN: A survey,” 2016 International Conference on ICT in Business Industry & Government (ICTBIG), Indore, 2016, pp. 1-6.