تحليل داده هاي ترافيکي با هدف تشخيص ازدحام با بهره گيري از الگوريتم هاي يادگيري ماشين
محورهای موضوعی : هوش مصنوعی و رباتیکناهید امانی 1 , محمدحسین عامری مهر 2 , سارا افاضاتی 3 , علی جاویدانی 4
1 - گروه مدیریت یکپارچه شبکه، پژوهشکده فناوری ارتباطات، پژوهشگاه ارتباطات و فناوری اطلاعات ، تهران، ایران
2 - گروه مدیریت یکپارچه شبکه، پژوهشکده فناوری ارتباطات، پژوهشگاه ارتباطات و فناوری اطلاعات ، تهران، ایران
3 - گروه مدیریت یکپارچه شبکه، پژوهشکده فناوری ارتباطات، پژوهشگاه ارتباطات و فناوری اطلاعات ، تهران، ایران
4 - دانشکده مهندسی برق و کامپیوتر، دانشکدگان فنی، دانشگاه تهران
کلید واژه: تشخیص ازدحام, يادگيري ماشين, XGBoost, معیارهای ارزیابی الگوریتم های یادگیری ماشین,
چکیده مقاله :
شناسايي ازدحام يکي از ارکان پايه اي در تضمين کيفيت خدمت به عنوان مهمترين معيار سنجش کارايي در شبکه هاي مخابراتي نسل جديد است. ازدحام منجر به از دست رفتن بسته هاي اطلاعاتي در شبکه مي شود که به دليل بالاتر بودن نرخ ارسال از ظرفيت در برخي از لينک‌هاي شبکه رخ مي دهد. براي کنترل ازدحام در شبکه گام نخست تمايز ميان از دست رفتن اطلاعات در اثر ازدحام و يا در اثر ساير موارد از جمله خرابي لينک است زيرا در صورتي که منشا از دست رفتن بسته به اشتباه ازدحام تشخيص داده شود کاهش نرخ ارسال در فرستنده کمکي به کاهش ازدحام نکرده و تنها موجب کاهش گذردهي و کيفيت سرويس مي شود. از اين رو مسئله اصلي اين مقاله شناسايي ازدحام و تفکیک آن از خطاي ناشي از لينک هاي ارتباطي در يک نمونه داده ترافيکي است. در اين مقاله براي حل مسئله مذکور از الگوريتم هاي مختلف يادگيري ماشين نظارتشده از جمله درخت تصميم، جنگل تصادفي، ماشين بردار پشتيبان، لجيستيک رگرسيون، K-نزديکترين همسايه، XGBoost، شبکه عصبي و تقويت درخت تصميم بهره گرفته شده است. الگوريتم هاي مذکور بر اساس معيارهاي مختلف از جمله دقت، درستی، F1-measure، حساسیت و AUC مورد ارزيابي قرار گرفته و با يکديگر مقايسه شده اند. اين ارزيابي بر اساس روش K-fold Cross Validation انجام شده است. نتایج شبیه‌سازی نشان می‌دهد که الگوریتم XGBoost به لحاظ تمام معیارهای ارزیابی نسبت به دیگر الگوریتم‌ها عملکرد بهتری در زمینه تشخیص ازدحام دارد.
Congestion detection is one of the basic elements in guaranteeing the quality of service and is one of the most important measures of efficiency in new-generation telecommunication networks. Congestion leads to the loss of information packets in the network because the sending rate is higher than the capacity in some network links. To control congestion in the network, the first step is to distinguish between the loss of packets due to congestion with the packet loss introduced by other cases, including link failure. Because if the source of the loss is mistakenly identified as congestion, reducing the sending rate in the transmitter does not mitigate the congestion and only reduces throughput and quality of service. Therefore, the main problem of this paper is to identify congestion and distinguish it from the error caused by communication links in a traffic data sample. In solving this problem, various supervised machine learning algorithms including decision tree, random forest, support vector machine, logistic regression, K-nearest neighbor, XGBoost, and neural network have been used. The mentioned algorithms have been evaluated based on different criteria and compared with each other.
[1] Z. Zhang; Y. Zhou; L. Teng; W. Sun; C. Li and X.Min, "Quality-of-Experience Evaluation for Digital Twins in 6G Network Environments," in IEEE Transactions on Broadcasting, pp. 1-13, 2024.
[2] R. Prasad and A. Sunny, "QoS-Aware Scheduling in 5G Wireless Base Stations," in IEEE/ACM Transactions on Networking, vol. 32, no. 3, pp. 1999-2011, June 2024.
[3] J. P. Astudillo León, L. J. Cruz Llopis and F. J. Rico-Novella, “A machine learning based Distributed Congestion Control Protocol for multi-hop wireless networks,” in Computer Networks, vol. 231, July 2023.
[4] J. Lorincz, Z. Klarin and J. Ožegovic, “A Comprehensive Overview of TCP Congestion Control in 5G Networks: Research Challenges and Future Perspectives,” in MDPI Sensors, vol. 21, no. 13, pp. 4510, 2021.
[5] L. Machora. "A survey of transmission control protocol variants," in World Journal of Advanced Research and Reviews, vol. 21, no. 3, pp. 1828– 1853, 2024.
[6] I. Stoica, "A Comparative Analysis of TCP Tahoe, Reno, New-Reno, SACK and Vegas," University of California Berkeley, CA, 2005.
[7] Z. Tafa, Zhilbert, and V. Milutinovic. "Machine learning in congestion control: A survey on selected algorithms and a new roadmap to their implementation." arXiv preprint arXiv:2112.15522 2021.
[8] T. Zhang and S. Mao, "Machine Learning for End-to-End Congestion Control," in IEEE Communications Magazine, vol. 58, no. 6, pp. 52-57, June 2020.
[9] Huiling Jiang, Qing Li, Yong Jiang, GengBiao Shen, Richard Sinnott, Chen Tian, Mingwei Xu, “When machine learning meets congestion control: A survey and comparison”, in Computer Networks, vol. 192, pp. 1-28, 2021.
[10] M. B. M. Kamel, I. Ahmed Najm and A. Khalaf Hamoud, "Congestion Control Prediction Model for 5G Environment Based on Supervised and Unsupervised Machine Learning Approach," in IEEE Access, vol. 12, pp. 91127-91139, 2024
[11] J. Alejandrino, R. Concepcion, S. Lauguico, M. G. Palconit, A. Bandala and E. Dadios, "Congestion Detection in Wireless Sensor Networks Based on Artificial Neural Network and Support Vector Machine," in 12th International Conf. on Humanoid, Nanotechnology, Inf. Tech., Commun. and Control, Environment, and Management (HNICEM), Manila, Philippines, 2020, pp. 1-6.
[12] P.T. Kasthuribai, “Optimized support vector machine based congestion control in wireless sensor network based Internet of Things,” in Int. J. Computer Network and Application, vol. 8, no. 4, pp. 444-454, 2021.
[13] N. Kumar, and M. Raubal. "Applications of deep learning in congestion detection, prediction and alleviation: A survey." in Transportation Research Part C: Emerging Technologies vol.133 pp. 103432, 2021.
[14] G. W. Geremew, , J. Ding. "Elephant Flows Detection Using Deep Neural Network, Convolutional Neural Network, Long Short‐Term Memory, and Autoencoder." in Journal of Computer Networks and Communications, pp. 1-18, 2023.
[15] M. Labonne, C. Chatzinakis and A. Olivereau, "Predicting Bandwidth Utilization on Network Links Using Machine Learning," in European Conf. on Networks and Commun. (EuCNC), Dubrovnik, Croatia, 2020, pp. 242-247.
[16] P. Geurts, I. El Khayat and G. Leduc, "A machine learning approach to improve congestion control over wireless computer networks," in Fourth IEEE International Conf. on Data Mining (ICDM'04), Brighton, UK, 2004, pp. 383-386.
[17] I. El Khayat, P. Geurts, and G. Leduc, ‘Improving TCP in Wireless Networks with an Adaptive Machine-Learnt Classifier of Packet Loss Causes’, in NETWORKING, 2005, pp. 549–560.
[18] J. D. Kelleher, B. M. Namee and A. D. Arcy. Fundamentals of machine learning for predictive data analytics: algorithms, worked examples, and case studies. MIT press, 2020.
[19] T. K. Ho, “Random Decision Forests,” in Proceedings of the 3rd International Conf. on Document Analysis and Recognition, Montreal, QC, 14–16 Aug., 1995, pp. 278–282.
[20] J. Tolles, WJ. Meurer, “Logistic Regression: Relating Patient Characteristics to Outcomes,” in JAMA, Vol. 316, no. 5, pp. 533-534, Aug. 2016.
[21] E. Fix, J. L. Hodges, “Discriminatory Analysis. Nonparametric Discrimination: Consistency Properties,” in International Statistical Review, vol. 57, no.3, pp. 238-247, 1989.
[22] R. O. Duda and P. E. Hart, Pattern classification. John Wiley & Sons, 2006.
[23] T. Chen, "Story and Lessons behind the evolution of XGBoost,” 2016.
[24] LeCun, Y. Bengio, and G. Hinton, "Deep learning," in Nature, vol. 521, no. 7553, pp. 436-444, 2015.
[25] D. Shi, W. Ping, A. Khushnood. "A survey on deep learning and its applications," in Computer Science Review, vol. 40, pp. 100379, 2021.
[26] Y. Freund, R. E Schapire. "A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting," in J. of Computer and System Sciences, vol. 55, no. 1, pp. 119-139, 1997.
[27] C. M. Bishop, and N. M. Nasrabadi. Pattern recognition and machine learning. Vol. 4. No. 4. New York: springer, 2006.
[28] [Online]. Available: https://people.montefiore.uliege.be/geurts/Papers/Boosting-DT/BD-Fifo.dat.gz