7 February 2023 in Expert Systems with Applications
6 August 2018 in Journal of Optical Communications and Networking
In this paper, we first review how the main machine learning concepts can apply to communication networks. Then we present results from a concrete application using unsupervised machine learning in a real network. We show how the application can detect anomalies at multiple network layers, including the optical layer, how it can be trained to anticipate anomalies before they become a problem, and how it can be trained to identify the root cause of each anomaly. Finally, we elaborate on the importance of this work and speculate about the future of intelligent adaptive networks.
26 July 2021 in IEEE Open Journal of Signal Processing
HDHossein Ebrahimi Dinaki
The end users’ satisfactory Quality of Experience (QoE) is a fundamental criterion for networked video service providers such as video-on-demand providers (Netflix, YouTube, etc.), cloud gaming providers (Google Stadia, PlayStation Now, etc.) and videoconferencing providers (Zoom, Microsoft Teams, etc.). To know the QoE, providers today typically predict it from the Quality of Service (QoS) parameters or the client-side's actual QoE metrics measured at the current time-step. But the former does not precisely reflect the users' experience, and the latter has a delay between QoE measurements at the client-side and the user's current experience. Mitigating this delay can provide a noticeable improvement in the delivery system's performance. For example, accurate forecasting of QoE for the near future allows the service management system to take a proactive approach and fix delivery issues before they become a noticeable problem at the end user, or at least reduce overall QoE degradation. QoE forecasting can also be used in rate adaptation in DASH or resource allocation in wireless networks. In this paper, we propose a method to prognosticate QoE metrics. Using data collected from an industry video streaming testbed for three different classes, we define a multivariate time series forecasting problem. We then model a hybrid state-of-the-art deep learning method, BiLSTM-CNN, to forecast the QoE metrics in advance. Evaluation of our proposed method compared to four other well-known ML models of Support Vector Regression (SVR), Multi-Layer Perceptron (MLP), Long short-term memory (LSTM), and Bidirectional LSTM (BiLSTM) demonstrates the superior performance of our proposed method.
5 July 2021 in IEEE Transactions on Instrumentation and Measurement
Although the autonomous detection of network status and localization of network faults can be a valuable tool for network and service operators, very few works have investigated this subject. As a result in today's networks, fault detection and localization remains a mostly manual process. In this article, we propose a machine learning (ML) method that can automatically detect the status of a network and localize faults. Our method uses the decision tree, gradient boosting (GB), and extreme GB ML algorithms to detect the network status as normal, congestion, and network fault. In comparison, existing related work can at best classify the network status as faulty or nonfaulty. Experimental results show that our method yields accuracies of up to 99% on a dataset collected through an emulated network.
26 August 2019 in Journal of Optical Communications and Networking
Maximizing the delivered capacity of optical transponders by mining the SNR margin to a near-zero level is of critical importance for the economic viability of future optical networks. We examine this trend from the equipment design point of view and apply artificial intelligence to the problem of capacity maximization. The bottom-up approach is focused on providing the key metrics that are built on the fundamentals of network performance, whereas the top-down approach uses machine learning and policy-driven actions that have the promise to achieve an unprecedented level of control. These methods show promise in delivering a practical near-zero margin network deployment.
12 July 2021 in Journal of Optical Communications and Networking
Loss of signal (LOS) represents a significant cost for operators of optical networks. By studying large sets of real-world performance monitoring data collected from six international optical networks, we find that it is possible to forecast LOS events with good precision one to seven days before they occur, albeit at relatively low recall, with supervised machine learning (ML). Our study covers 12 facility types, including 100G lines and ETH10G clients. We show that the precision for a given network improves when training on multiple networks simultaneously relative to training on an individual network. Furthermore, we show that it is possible to forecast LOS from all facility types and all networks with a single model, whereas fine-tuning for a particular facility or network brings only modest improvements. Hence our ML models remain effective for optical networks previously unknown to the model, which makes them usable for commercial applications.