Haytham Assem

Monitoring VoIP call quality using improved simplified E-model

ITU-T recommendation G.107 introduced the E-model, a repeatable way to assess if a network is prepared to carry a VoIP call or not. Various studies show that the E-model is complex with many factors to be used in monitoring purposes. Consequently, simplified versions of the E-model have been proposed to simplify the calculations and focus on the most important factors required for monitoring the call quality. In this paper, we propose simple correction to a simplified E-model; we show how to calculate the correction coefficients for 4 common codecs (G.711, G.723.1, G.726 and G.729A) and then we show that its predictions better match PESQ scores by implementing it in a monitoring application.

CogNet: A network management architecture featuring cognitive capabilities

It is expected that the fifth generation mobile networks (5G) will support both human-to-human and machine-to-machine communications, connecting up to trillions of devices and reaching formidable levels of complexity and traffic volume. This brings a new set of challenges for managing the network due to the diversity and the sheer size of the network. It will be necessary for the network to largely manage itself and deal with organisation, configuration, security, and optimisation issues. This paper proposes an architecture of an autonomic self-managing network based on Network Function Virtualization, which is capable of achieving or balancing objectives such as high QoS, low energy usage and operational efficiency. The main novelty of the architecture is the Cognitive Smart Engine introduced to enable Machine Learning, particularly (near) real-time learning, in order to dynamically adapt resources to the immediate requirements of the virtual network functions, while minimizing performance degradations to fulfill SLA requirements. This architecture is built within the CogNet European Horizon 2020 project, which refers to Cognitive Networks.

Can machine learning aid in delivering new use cases and scenarios in 5G

5G represents the next generation of communication networks and services, and will bring a new set of use cases and scenarios. These in turn will address a new set of challenges from the network and service management perspective, such as network traffic and resource management, big data management and energy efficiency. Consequently, novel techniques and strategies are required to address these challenges in a smarter way. In this paper, we present the limitations of the current network and service management and describe in detail the challenges that 5G is expected to face from a management perspective. The main contribution of this paper is presenting a set of use cases and scenarios of 5G in which machine learning can aid in addressing their management challenges. It is expected that machine learning can provide a higher and more intelligent level of monitoring and management of networks and applications, improve operational efficiencies and facilitate the requirements of the future 5G network.

Improved E-model for monitoring quality of multi-party VoIP communications

Maintaining good Quality-of-Experience (QoE) is crucial for Voice-over-IP (VoIP) applications, particularly those operating across the public Internet. Accurate online estimation of QoE as perceived by end users allows VoIP applications take steps to improve QoE when it falls below acceptable levels. ITU-T recommendation G.107 introduced the E-model, which provides a means to assess QoE levels for two-party VoIP sessions. In this paper we provide an analysis of the accuracy of the E-model for multi-party VoIP sessions when all audio is processed by a centralised focus node. We analyse the impact of what we term the “Focus Transcoding Effect (FTE),” the “Focus Forwarding Effect (FFE),” and the number of end-points participating in the session. Through comparison to QoE metrics produced by the offline PESQ method for three common audio codecs, we show that the standard E-model does not provide accurate QoE assessment for multi-party VoIP sessions. We then introduce an improved Emodel for these codecs for multi-party VoIP sessions. We describe the implementation of the improved E-model in a QoE monitoring application, showing that it produces results similar to actual PESQ scores.

Towards Bridging the Gap between Machine Learning Researchers and Practitioners

As data keeps growing, Big Data starts to be everywhere, and there is almost an urgent need to make sense of this data. This is why Machine Learning has become crucial as it aids in improving business, decision making and it has the potential to provide solutions for a wide range of problems in computer science and other fields. Machine Learning (a.k.a. Data Mining or Predictive Analytics) algorithms can learn how to perform certain tasks by generalizing from the out of sample examples. This is a totally different paradigm than traditional programming language approaches based on writing programs that process data to produce an output. However, choosing a suitable machine learning algorithm for a particular application requires substantial amount of effort that is even hard to undertake even with text books. In order to reduce the effort, this paper introduces a recommender system that will aid machine learning researchers and practitioners to choose the optimum machine learning model to use. The system is based on an approach that is introduced in the paper called TCDC which stands for Train, Compare, Decide, and Change.

ADE: An ensemble approach for early Anomaly Detection

Proactive anomaly detection refers to anticipating anomalies or abnormal patterns within a dataset in a timely manner. Discovering anomalies such as failures or degradations before their occurrence can lead to great benefits such as the ability to avoid the anomaly happening by applying some corrective measures in advance (e.g., allocating more resources for a nearly saturated system in a data centre). In this paper we address the proactive anomaly detection problem through machine learning and in particular ensemble learning. We propose an early Anomaly Detection Ensemble approach, ADE, which combines results of state-of-the-art anomaly detection techniques in order to provide more accurate results than each single technique. Moreover, we utilise a a weighted anomaly window as ground truth for training the model, which prioritises early detection in order to discover anomalies in a timely manner. Various strategies are explored for generating ground truth windows. Results show that ADE shows improvements of at least 10% in earliest detection score compared to each individual technique across all datasets considered. The technique proposed detected anomalies in advance up to ∼16h before they actually occurred.

RCMC: Recognizing Crowd-Mobility Patterns in Cities Based on Location Based Social Networks Data

During the past few years, the analysis of data generated from Location-Based Social Networks (LBSNs) have aided in the identification of urban patterns, understanding activity behaviours in urban areas, as well as producing novel recommender systems that facilitate users’ choices. Recognizing crowd-mobility patterns in cities is very important for public safety, traffic managment, disaster management, and urban planning. In this article, we propose a framework for Recognizing the Crowd Mobility Patterns in Cities using LBSN data. Our proposed framework comprises four main components: data gathering, recurrent crowd-mobility patterns extraction, temporal functional regions detection, and visualization component. More specifically, we employ a novel approach based on Non-negative Matrix Factorization and Gaussian Kernel Density Estimation for extracting the recurrent crowd-mobility patterns in cities illustrating how crowd shifts from one area to another during each day across various time slots. Moreover, the framework employs a hierarchical clustering-based algorithm for identifying what we refer to as temporal functional regions by modeling functional areas taking into account temporal variation by means of check-ins’ categories. We build the framework using a spatial-temporal dataset crawled from Twitter for two entire years (2013 and 2014) for the area of Manhattan in New York City. We perform a detailed analysis of the extracted crowd patterns with an exploratory visualization showing that our proposed approach can identify clearly obvious mobility patterns that recur over time and location in the urban scenario. Using same time interval, we show that correlating the temporal functional regions with the recognized recurrent crowd-mobility patterns can yield to a deeper understanding of city dynamics and the motivation behind the crowd mobility. We are confident that our proposed framework not only can help in managing complex city environments and better allocation of resources based on the expected crowd mobility and temporal functional regions but also can have a direct implication on a variety of applications such as personalized recommender systems, anomalous event detection, disaster resilience management systems, and others.

A new adaptive redundancy control algorithm for VoIP applications

Packet loss is one of the most important factors in degrading Voice over IP (VoIP) perceived call quality. Forward Error Correction (FEC) is a powerful technique for transmitting audio streams over the IP network to decrease the effect of packet loss. Although these method reduces the effect of packet loss, it increases the bandwidth and delay in order to recover from the lost packets. In this paper, we propose a new adaptive FEC mechanism for voice calls based on the generated codewords from a Reed-Solomon (RS) encoder. This mechanism chooses the optimum RS code from a family of codes to improve the conversational call quality. Our proposed mechanism is able to switch between different codes during the call to account for the variation of the network conditions including packet loss and delay. We have deduced the proposed algorithm by performing subjective mean opinion score (MOS) testing based on an interactive assessment tests. We show that our adaptive algorithm outperforms fixed RS codes under highly varying network conditions.

DeepAD: A Generic Framework Based on Deep Learning for Time Series Anomaly Detection

This paper presents a generic anomaly detection approach for time-series data. Existing anomaly detection approaches have several drawbacks such as a large number of false positives, parameters tuning difficulties, the need for a labeled dataset for training, use-case restrictions, or difficulty of use. We propose DeepAD, an anomaly detection framework that leverages a plethora of time-series forecasting models in order to detect anomalies more accurately, irrespective of the underlying complex patterns to be learnt. Our solution does not rely on the labels of the anomalous class for training the model, nor for optimizing the threshold based on highest detection given the labels in the training data. We compare our framework against EGADS framework on real and synthetic data with varying time-series characteristics. Results show significant improvements on average of 25% and up to \(40-50\)% in \(F_1{\text{- }}score\), precision, and recall on the Yahoo Webscope Benchmark.

Urban Water Flow and Water Level Prediction Based on Deep Learning

The future planning, management and prediction of water demand and usage should be preceded by long-term variation analysis for related parameters in order to enhance the process of developing new scenarios whether for surface-water or ground-water resources. This paper aims to provide an appropriate methodology for long-term prediction for the water flow and water level parameters of the Shannon river in Ireland over a 30-year period from 1983–2013 through a framework that is composed of three phases: city wide scale analytics, data fusion, and domain knowledge data analytics phase which is the main focus of the paper that employs a machine learning model based on deep convolutional neural networks (DeepCNNs). We test our proposed deep learning model on three different water stations across the Shannon river and show it out-performs four well-known time-series forecasting models. We finally show how the proposed model simulate the predicted water flow and water level from 2013–2080. Our proposed solution can be very useful for the water authorities for better planning the future allocation of water resources among competing users such as agriculture, demotic and power stations. In addition, it can be used for capturing abnormalities by setting and comparing thresholds to the predicted water flow and water level.