物聯網(Internet of Things，IoT)在現今的社會中扮演著越來越重要的角色，不管是在製造業、農業抑或是在家庭中，都能看到物聯網對於生活品質或是生產效率的提升。但也因為物聯網是由多種的資訊技術整合在一起的，具有大規模以及異構的性質，容易產生許多的資訊安全漏洞。攻擊者可以利用這些漏洞執行惡意操作，進而導致使用者的損失。因此，準確又有效率的資安防護方法對於物聯網服務是非常重要的。
近十年來，深度學習模型在影像辨識領域上大放異彩，像是應用在自動駕駛車輛的物件辨識，以及用於身分認證的人臉辨識，其中的關鍵就是卷積神經網路(Convolutional Neural Network, CNN)的蓬勃發展。近年來有研究提出深度可分離卷積[ 22 ]，可以在大量減少運算資源的情況下，達到與原始卷積神經網路同樣的效能。在深度可分離卷積中較著名的架構為Xception模組，因為該架構在知名的ImageNet 資料集上勝過所有過去研究的效能。
本研究提出了一個基於Xception模組的深度學習模型，預測是否有物聯網攻擊的發生並判斷該攻擊的種類。最終本研究提出的模型在效能上勝過過去的研究，分別在準確率與F1-score達到99%與99.2%。本研究除了提出一個效能更好的深度學型模型外，也進一步對分散式阻斷服務攻擊(Distributed Denial-of-Service, DDoS)進行案例分析並找出本研究模型在使用何種特徵上更具有優勢。

The Internet of Things (IoT) is playing an increasingly important role in today's society. Whether it is in manufacturing, agriculture or house, the Internet of Things can improve the quality of life or production efficiency. However, because the Internet of Things is integrated by a variety of information technologies, with a large-scale and heterogeneous nature, it is prone to produce many information security vulnerabilities. Attackers can exploit these vulnerabilities to execute malicious operations, which in turn lead to losses for users. Therefore, accurate and efficient information security protection methods are very important for IoT services.
In the past 10 years, deep learning models have been brilliant in the field of image recognition, such as object recognition in self-driving vehicles, and face recognition for identity authentication. The key to this is the vigorous development of Convolutional Neural Network (CNN). In recent years, some studies have proposed depthwise separable convolution [ 22 ], which can achieve the same performance as the original convolutional neural network with a large reduction in computing resources. The well-known architecture in depthwise separable convolution is the Xception module, as this architecture outperforms all past studies on the famous ImageNet dataset.
This study proposes a deep learning model based on the Xception module to predict whether there is an IoT attack and determine the type of the attack. Finally, the model proposed in this study outperformed previous studies in terms of performance, reaching 99% and 99.2% in accuracy and F1-score, respectively. In addition to proposing a more efficient deep learning model, this study further conducts a case study on Distributed Denial-of-Service (DDoS) attacks. And find out in which features the model in this study is more advantageous.

[ 1 ] He, K., Zhang, X., Ren, S., & Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 770-778). (2016).
[ 2 ] Alsaedi, A., Moustafa, N., Tari, Z., Mahmood, A., & Anwar, A. TON_IoT telemetry dataset: A new generation dataset of IoT and IIoT for data-driven intrusion detection systems. IEEE Access, 8, 165130-165150. (2020).
[ 3 ] Abdel-Basset, M., Hawash, H., & Sallam, K. Federated threat-hunting approach for microservice-based industrial cyber-physical system. IEEE Transactions on Industrial Informatics, 18(3), 1905-1917 (2021).
[ 4 ] Kolias, C., Kambourakis, G., Stavrou, A., & Gritzalis, S. Intrusion detection in 802.11 networks: empirical evaluation of threats and a public dataset. IEEE Communications Surveys & Tutorials, 18(1), 184-208. (2015).
[ 5 ] Zargar, S. T., Joshi, J., & Tipper, D. A survey of defense mechanisms against distributed denial of service (DDoS) flooding attacks. IEEE communications surveys & tutorials, 15(4), 2046-2069. (2013).
[ 6 ] Krupp, J., Backes, M., & Rossow, C. Identifying the scan and attack infrastructures behind amplification DDoS attacks. In Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security (pp. 1426-1437). (2016, October).
[ 7 ] Nelson, T., & Chaffin, M. Common cybersecurity vulnerabilities in industrial control systems. Control systems security program. (2011).
[ 8 ] Al-rimy, B. A. S., Maarof, M. A., & Shaid, S. Z. M. Ransomware threat success factors, taxonomy, and countermeasures: A survey and research directions. Computers & Security, 74, 144-166. (2018).  
[ 9 ] Zolanvari, M., Teixeira, M. A., Gupta, L., Khan, K. M., & Jain, R. Machine learning-based network vulnerability analysis of industrial Internet of Things. IEEE Internet of Things Journal, 6(4), 6822-6834. (2019).
[ 10 ] Zargar, S. T., Joshi, J., & Tipper, D. A survey of defense mechanisms against distributed denial of service (DDoS) flooding attacks. IEEE communications surveys & tutorials, 15(4), 2046-2069. (2013).
[ 11 ] He, H., Bai, Y., Garcia, E. A., & Li, S. ADASYN: Adaptive synthetic sampling approach for imbalanced learning. In 2008 IEEE international joint conference on neural networks (IEEE world congress on computational intelligence) (pp. 1322-1328). IEEE. (2008, June).
[ 12 ] McCulloch, W. S., & Pitts, W. A logical calculus of the ideas immanent in nervous activity. The bulletin of mathematical biophysics, 5(4), 115-133. (1943).
[ 13 ] Hussain, B., Du, Q., Sun, B., & Han, Z. Deep learning-based DDoS-attack detection for cyber–physical system over 5G network. IEEE Transactions on Industrial Informatics, 17(2), 860-870. (2020).
[ 14 ] Liu, Y., Garg, S., Nie, J., Zhang, Y., Xiong, Z., Kang, J., & Hossain, M. S. Deep anomaly detection for time-series data in industrial IoT: A communication-efficient on-device federated learning approach. IEEE Internet of Things Journal, 8(8), 6348-6358. (2020).
[ 15 ] Li, B., Wu, Y., Song, J., Lu, R., Li, T., & Zhao, L. DeepFed: Federated deep learning for intrusion detection in industrial cyber–physical systems. IEEE Transactions on Industrial Informatics, 17(8), 5615-5624. (2020).
[ 16 ] Hochreiter, S., & Schmidhuber, J. Long short-term memory. Neural computation, 9(8), 1735-1780. (1997).
[ 17 ] Sifre, L., & Mallat, S. Rigid-motion scattering for texture classification. arXiv preprint arXiv:1403.1687. (2014).
[ 18 ] Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., & Wojna, Z. Rethinking the inception architecture for computer vision. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 2818-2826). (2016).
[ 19 ] Chollet, F. Xception: Deep learning with depthwise separable convolutions. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 1251-1258). (2017).
[ 20 ] Wu, Y., & He, K. Group normalization. In Proceedings of the European conference on computer vision (ECCV) (pp. 3-19). (2018).
[ 21 ] Yu, X., Li, M., Kim, T., Jeong, S. P., & Ryu, K. H. An application of improved gap-BIDE algorithm for discovering access patterns. Applied Computational Intelligence and Soft Computing, 2012.
[ 22 ] Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., ... & Adam, H. Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861. (2017).