Perceived quality prediction models for multimedia services vary greatly depending on the type of the data and on the amount of information related to the original signal used. In this research, we have developed machine learning-based reduced-reference bitstream audiovisual quality prediction models by using the parametric version of the publicly available INRS audiovisual quality dataset. As that original INRS dataset did not contain bitstream information but provided both reference and transmitted videos, we have computed its bitstream version to develop the reduced-reference bitstream models. We have compared the performance of the Decision Trees based ensemble methods, Genetic Programming and Deep Learning models on this bitstream version of the dataset and have also compared these results with the results of the no-reference parametric models on the parametric version of the dataset. Decision Trees based ensemble methods outperformed Deep Learning and Genetic Programming based models when reduced-reference bitstream data was used and outperformed all existing no-reference parametric models that were trained and tested on the parametric version of the dataset. Our studies show that Decision Trees based approaches are well suited for no-reference parametric models as well as for reduced-reference bitstream models.

In order to mechanically predict audiovisual quality in interactive multimedia services, we have developed machine learning--based no-reference parametric models. We have compared Decision Trees--based ensemble methods, Genetic Programming and Deep Learning models that have one and more hidden layers. We have used the Institut national de la recherche scientifique (INRS) audiovisual quality dataset specifically designed to include ranges of parameters and degradations typically seen in real-time communications. Decision Trees--based ensemble methods have outperformed both Deep Learning-- and Genetic Programming--based models in terms of Root-Mean-Square Error (RMSE) and Pearson correlation values. We have also trained and developed models on various publicly available datasets and have compared our results with those of these original models. Our studies show that Random Forests--based prediction models achieve high accuracy for both the INRS audiovisual quality dataset and other publicly available comparable datasets.

We have developed reduced reference parametric models for estimating perceived quality in audiovisual multimedia services. We have created 144 unique configurations for audiovisual content including various application and network parameters such as bitrates and distortions in terms of bandwidth, packet loss rate and jitter. To generate the data needed for model training and validation we have tasked 24 subjects, in a controlled environment, to rate the overall audiovisual quality on the absolute category rating (ACR) 5-level quality scale. We have developed models using Random Forest and Neural Network based machine learning methods in order to estimate Mean Opinion Scores (MOS) values. We have used information retrieved from the packet headers and side information provided as network parameters for model training. Random Forest based models have performed better in terms of Root Mean Square Error (RMSE) and Pearson correlation coefficient. The side information proved to be very effective in developing the model. We have found that, while the model performance might be improved by replacing the side information with more accurate bit stream level measurements, they are performing well in estimating perceived quality in audiovisual multimedia services.

We present the INRS audiovisual quality dataset made of 160 unique configurations for audiovisual content including various media compression and network distortion parameters such as video frame rate, quantization and noise reduction parameters, and packet loss rate. The compression and network distortion parameter range values are selected to match real-time communications use cases. The H.264 video codec in 720p resolution and the AMR-WB audio codec are used for encoding video and audio streams. Thirty observers have rated the overall audiovisual quality on the Absolute Category Rating (ACR) 5-level quality scale in a controlled environment. The dataset includes MOS values, packet loss rates measured at bit stream level for both video and audio streams, compression parameters and various packet header information. We have used open source software for producing source audiovisual sequences, end-to-end streaming and a custom video player. These tools and the dataset are free to public access for research and development purposes.

“Over the top” (OTT) has become an increasingly popular form of delivery for voice and video communications. Ubiquitous OTT communications, pioneered by Skype, have, with the recent release of WebRTC, become a commodity readily available for various applications, without a need for extra installations. Still, in spite of such progress, media communications remain challenging as user expectations are keeping ahead of technology progress and portable devices are becoming the norm. In practical terms, this means that media flows must constantly adapt to varying network conditions to try to -reliably!- offer the best experience to the users. We present in this paper a sender-side adaptation algorithm for video flows which trades off redundancy and throughput to maximize quality while minimizing the likelihood of loss, thus preserving service continuity. By exploiting standard FEC mechanisms at different levels of redundancy while controlling the quality of media delivered, we create a discrete parameter space which the algorithm exploits based on quality feedback. We study the performance of this algorithm in a number of reference test cases and discuss how it improves on earlier and related proposals.

Content Delivery Networks have become commonly used by content providers such as Netflix to deliver TV programmes to a large number of customers. This solution however breaks down when intensive processing is required and when content cannot be stored ahead of time, as is the case with live (real time) content. We study here resource management issues when a cloud infrastructure is used to deliver live content. More specifically, we look at the allocation of machines to host servers and how to keep that number as small as possible while meeting user demand. We express the problem as an online bin packing model and show how, by exploiting a simple property of the media, we can derive an efficient solution.

Wireless mesh networks (WMN) are efficient and low cost solutions for the deployment of broadband access in various environments. While the capacity and the bandwidth availability of single radio WMNs may severely limit QoS, multiradio cognitive WMNs (CWMN) can provide better QoS mechanisms. This paper describes a multihop backpressure scheduling algorithm which ensures that the available bandwidth is properly shared considering the type of traffic to forward, the distance of the clients from the portal and the fluctuating conditions of the links. Performance evaluation in our simulated environment shows that, while preventing bandwidth wastage, our new scheduling algorithm achieves nearly perfect fairness.

ABSTRACT We present an original approach to identify synchronize (SYN) flooding attacks from the victim's side, on the basis of a classification of the different forms that TCP handshakes can take during a connection set‐up between a client and a server (e.g. for Web traffic). We first identify the unusual handshake sequences that result from an attack and show how such observations can be used for SYN flooding attack detection. We then introduce a data structure to monitor, in real time, the state of the TCP handshake and study its performance. In addition, we explain the management of the data structure for operations such as initialization, adding and removing flows. Finally, we analyse the effectiveness of our TCP handshake monitoring to identify the presence of SYN flooding attacks by applying it to real traffic traces. To allow quick protection and help guarantee a proper defence, the detection is done in real time. Our detection system uses a non‐parametric cumulative sum algorithm (CUSUM), which has the benefit of not requiring a detailed model of the normal and attack traffic while achieving excellent detection levels. Copyright © 2011 John Wiley & Sons, Ltd.

Wireless mesh networks (WMN) are efficient and low cost solutions for the deployment of broadband access in various environments. To support real-time applications such as multimedia and emergency services, WMNs must provide quality of service (QoS) guarantees and QoS continuity protection. While the capacity of single radio WMNs may severely limit the QoS for such traffic, multiradio WMNs can overcome this restriction and provide additional links to support better QoS mechanisms. This paper describes two self-configuration mechanisms to establish mesh topologies in IEEE 802.11s multiradio WMNs: a distributed mechanism based on the portal announcement protocol specified in the IEEE 802.11s standard to determine the initial channel assignment of the mesh points forming a backhaul network and a centralized mechanism to optimize the topology initially created once the network is stable. We also study the performance of topology formation in IEEE 802.11s multiradio WMNs and identify the various factors affecting delays in the creation and modification of topologies.

TCP (Transmission Control Protocol) is the dominant end to end transport protocol of the Internet, with a wide range of applications including Web, mail or peer to peer traffic. The TCP stack implements a “backlog queue” for new connections, which contains an entry for every client's connection setup received by the server. If the TCP handshake is not completed, the pending half-open connection stays in the backlog queue until a time-out expires and, if that time-out value is too big, the half-open connection stays in the queue longer than necessary. We present a technique to assign and find a suitable connection-establishment time-out value to reduce the risks of an overflow of the backlog queue in situations of SYN flooding attacks. We evaluate from experimental traces that our technique can reduce the size of the backlog queue size up to 50% while preserving normal connections.

Channel quality prediction is important for many networking mechanisms. This paper considers long-term prediction with the prediction distance of some hundred milliseconds. The prediction works in the context of Adaptive Coding and Modulation, which actually converts the task of continuous value prediction into discrete value prediction. In this work, pattern management is also considered as a key mechanism to reduce memory and processing overheads. Performance evaluation is carried out using measurement traces to evaluate the prediction algorithm with different parameter settings and matching measures.

We present a performance study of a wireless mesh network based on off-the-shelf 802.11 technology. Unlike earlier work, we explore the practical constraints of the use of a “vanilla”, unmodified 802.11 MAC layer. We also consider the use of multiple radios, in the 5GHz band, to reduce interference. We establish the origin of degradation and limits in performance of practical deployments and study means to circumvent them. We focus our analysis on simple means, deployed in the network without attempting to modify the Internet Protocol stack on clients and focusing on ubiquitous layer 3 mechanisms, which limit bandwidth and only influence sharing. Using as a yardstick equity between TCP flows, which more than ever represent the dominant support for application traffic in the Internet, we show which improvements can be achieved over unmanaged usage and at what cost, and discuss implications for their use.

We present an original approach to detect SYN flooding attacks from the victim's side, by monitoring unusual handshake sequences. Detection is done in real-time to allow quick protection and help guarantee a proper defence. Our detection system uses an entropy measure to detect changes in the balance of TCP handshakes. Experiment results show that our method can detect SYN flooding attacks with better accuracy and robustness than traditional stateless methods, and with manageable overhead.

A variety of wireless access technologies provides todaypsilas users with pervasive access to critical content and applications. Guaranteeing a satisfactory quality of service requires reliable and ubiquitous delivery schemes that optimally use the available access technologies. This paper presents a novel architecture to support adaptive real-time data and multimedia applications over different wireless technologies. It addresses several issues associated with the satisfactory operation of applications running over heterogeneous networks. The system is validated using a real-life healthcare application capable of transmitting voice, video and medical data simultaneously. Experimental results are reported that demonstrate the viability of the approach.

3G networks offer new opportunities for mobile or nomadic users to remain connected to their services through changes in environment conditions. Such continuity still presents challenges however. 3G networks require resource management to guarantee that services will meet performance expectations. Also, because of cost, alternate wireless networking technologies can now be used in complement with 3G to take advantage of hotspots, while still preserving continuity of session.In this paper, we propose and analyse a multi-network session management infrastructure based on established protocols. The session unifies information on active services over different access networks. It supports service continuity, migration of services between networks to attempt to always give to the user a better experience at lower cost.A session supports composition of services based on different signalling models, although our primary focus is on SIP and HTTP, and our concept derived from IMS.

In wireless mobile environments, where resources are scarce, the basis for QoS provisioning is to control the admission of new and handoff subscriber services, to avoid future detriment perturbations of established connections. Most research efforts considered only the bandwidth of the wireless link to control the admission of new services. In addition, proposals for degrading and enhancing services only account for mobility. In this paper, we propose a novel approach for service admission and adaptation that considers, in addition to the station mobility, both bandwidth and error rate requirements as well as the possibility of multiple services per mobile station, while minimizing interference to neighbour cells as well as signalling overhead.When compared to other proposals in a simulated 3G cellular system, our admission control approach outperformed by 7% the equal priority and guard capacities schemes in terms of total number of accepted services in the cell, especially in high handoff and new call rates. Moreover, our adaptation scheme succeeded in improving the admission probability by a further 6% in indoor environments and 9% in outdoor environments.

This paper develops a new network provisioning and resource allocation scheme. We introduce the concept of the effective burstiness curve (EBC), which is defined as a percentile of the maximum burstiness curve. For a fixed service rate, EBC represents the size of a buffer for which the probability of buffer overflow is arbitrarily small. We show that EBC is a convex nonincreasing function of the service rate. We also introduce the empirical effective burstiness curve (EEBC), an estimator of EBC, which can be obtained with a water-filling algorithm. For discrete queue size, EEBC can be evaluated with a recursive algorithm. The technique is applied to MPEG4 encoded video traces.

Packet delay traces are important sources of measurements for analyzing end-to-end performance of computer networks. Due to the lack of tight synchronization between the clocks of end systems, these measurements can be quite inaccurate. Therefore, detection, estimation and removal of clock skew from delay traces is a critical operation to obtain precise measurements of network latencies. In this paper, we propose two new techniques to detect, estimate and remove the clock skew in delay traces. The first technique, named average technique, derives the clock skew estimate by calculating the average of the difference between consecutive packets' delay. The second technique, named direct skew removal technique, proceeds by iteratively evaluating a set of possible skew values until the best value is reached. Compared with existing techniques such as linear programming and convex-hull, the average technique reduces the complexity of the skew estimation operation. The direct skew removal is more accurate and allows us to obtain delay after skew removal with the same precision as the original traces. Clock skew is also present in online delay measurements that are used by real-time endpoints such as audio and video terminals, to determine the buffering delay of received packets and to synchronize streams from different sources. The problem is more obvious when communication sessions last for a long time. Contrary to the offline skew removal, little work has been done on the online skew removal problem. We propose two simple algorithms to remove the clock skew from online delay measurements. The first algorithm, named the sliding window algorithm, tracks the skew by continually evaluating the variation of the minimum measured delay. The second algorithm, named the combined algorithm, is a mixed approach of the sliding window and the convex-hull algorithm proposed elsewhere.