Tools
Tools
Imtiaz, Nouman, Wahid, Abdul, Abideen, Syed Zain Ul, Kamal, Mian Muhammad, Sehito, Nabila, Khan, Salahuddin, Virdee, Bal Singh, Kouhalvandi, Lida and Alibakhshikenari, Mohammad (2025) A deep learning-based approach for the detection of various Internet of Things intrusion attacks through optical networks. Photonics, 12(1) (35). pp. 1-39. ISSN 2304-6732
The widespread use of the Internet of Things (IoT) has led to significant breakthroughs in several fields but has also caused a sharp increase in cybersecurity risks. This research introduces XIoT, a novel Explainable IoT attack detection model created to address the changing cyber risks facing IoT networks, particularly as they interact with optical communication infrastructure. XIoT utilizes advanced deep learning (DL) methods, namely convolutional neural networks (CNNs), to examine spectrogram pictures created from IoT network traffic data. By scrutinizing these images’ spatial and sequential aspects, XIoT offers a comprehensive and nuanced understanding of the underlying features of malicious activities. A key distinguishing feature of XIoT lies in its emphasis on interpretability, enabling stakeholders to gain insights into the rationale behind its predictions. By integrating explainable AI mechanisms, XIoT delivers accurate classifications of diverse IoT attacks and elucidates the key factors driving its decision-making process. This transparency enhances trust in the model’s outputs and facilitates informed decision-making by cybersecurity analysts and network administrators. Additionally, leveraging the high-speed, low-latency characteristics of optical networks, XIoT’s model can process extensive IoT data streams more efficiently, supporting real-time detection in expansive IoT ecosystems. To evaluate the efficacy of XIoT, comprehensive experiments are conducted on benchmark datasets encompassing a range of IoT attack scenarios. The datasets include KDD CUP99, UNSW NB15, and Bot-IoT, each representing distinct challenges and complexities inherent in IoT security. XIoT has exceptional accuracy rates of 99.34%, 99.61%, and 99.21% across various datasets, showcasing its resilience and adaptability in different IoT settings. XIoT outperforms current intrusion detection methods regarding accuracy and interpretability, as shown in comparison evaluations. XIoT is a powerful solution for protecting IoT ecosystems from cyber assaults by surpassing current models and offering practical insights on identified dangers.
Available under License Creative Commons Attribution 4.0.
Download (7MB) | Preview
 |
View Item |