A major driving viewpoint in the development of the Next Generation Networks concerns the importance of the user/customer needs in the provisioning of the requested service. Indeed, according to this view is the customer subscription and the customer requested service that dictate the running of the network, that is, upon receiving a customer request for a service, the network will assign the service quality and service priority according to the customer subscription and the requested service, and the service priority is used for traffic engineering in case of network congestion. Under such a customer-centric approach, the management plays a role as important as the technology. It is a revolutionary step to transfer from technology-centred approach to a customer-centred approach. According to this vision, the Quality of Service (QoS) is embraced by the Quality of Experience (QoE) concept, which introduces key issues in the design, deployment, provisioning and maintenance of future (and current) NGN architectures and services.
The Net4U Lab investigates these issues in the following areas:
- Subjective Video Quality Assessment – The objective of this study is to assess the subjective characteristics of video quality. It is concerned with how video is perceived by a viewer and designates his opinion on a particular video sequence, typically corrupted by one or more coding and/or transmission distortions. We conducted studies aimed at assessing the Quality of Experience (QoE) in video streaming when the user is using mobile devices (i.e. tablets). From this study we created a database containing subjective assessment scores relative to hundreds of streaming sessions of video sequences encoded with H.264/AVC and affected by distorsions (e.g., playout delay, transmission errors, buffer overflows). Subjective evaluations were conducted in compliance with the ITU-T Recommendation P.910 through single-stimulus Absolute Category Rating (ACR).
- 4K UHDTV Subjective Video Quality Assesments – Recent years have been characterized by the new so-called 4K format, also known as 2160p, which is part of the Ultra-high Definition television (UHDTV) standard defined by ITU-R BT.2020. ITU-R BT series define general laboratory and home viewing conditions with particular emphasis in PVD valid for both SDTV and HDTV whereas subjective video quality (SVQ) assessments for UHDTV are actually fully or partially missing. Usually, SVQ test evaluators rate their opinions about each video on paper, and then the data are manually managed through a computer system for further analysis. For these reasons, the new platform adds laboratory and home conditions valid for UHDTV (i.e., PVD, angle view, number of evaluators at a time, etc), simplifies the SVQ tests and automates the collection of data. MOS vs pMOS comparison for 4K video sequences at different bitrate is evaluated in terms of sex, age, and daily number of hours in front of TV.
- QoE prediction from face expressions and gaze direction – The objective of this study is the investigation on the potentials to implicitly estimate the Quality of Experience (QoE) of an user of video streaming services by acquiring a video of her face and monitoring her facial expression and gaze direction. To this, we conducted a crowdsourcing test in which participants were asked to watch and rate the quality when watching 20 videos subject to different impairments, while their face was recorded with their PC’s webcam. The following features were then considered: the Action Units (AU) that represent the facial expression, and the position of the eyes’ pupil that represent the gaze direction. The dataset containing these features, together with the provided quality rates collected from a total of 400 subject tests, is available at the following link. To get the password please send an email to firstname.lastname@example.org with the following subject: “Password request for QoE-FaceExpressions dataset”.
- GAN Generated Images for training Facial Expression Recognition systems – Most of Facial Expression Recognition (FER) systems rely on machine learning approaches that require large databases (DBs) for an effective training. As these are not easily available, a good solution is to augment the DBs with appropriate techniques, which are typically based on either geometric transformation or deep learning based technologies (e.g., Generative Adversarial Networks (GANs)). Whereas the first category of techniques have been fairly adopted in the past, studies that use GAN-based techniques are limited for FER systems. To advance in this respect, we evaluate the impact of the GAN techniques by creating a new DB containing the generated synthetic images. The face images contained in the KDEF DB are used as the base to create novel synthetic images using the facial features of 2 images selected from the YouTube-Faces DB. The dataset containing these GAN generated images is available at the following link.
- Enhanced QoE in Home Entertainment system – Net4U is currently carrying on research on Multimedia Broadcasting TV Streams (such as DVB, ATSC, or equivalent) supplementary side information to be used to enrich the QoE of the user. The idea is to connect the Smart TV to others Cyberg-Physical Sustem (CPS) devices by Internet of Things (IoT) within the home able to “react” opportunely when driven by this side information during the watching time. A CPS is a system of collaborating computational elements controlling physical entities. The growth in the number of smart devices and sensors (i.e., CPS) connected to the IoT has the potential to change how consumers interact with all networked technology, including their media and entertainment platforms. This represents an opportunity for the entertainment industry to assimilate the growing volume of customer insight that will be constantly generated by IoT technologies throughout the market in order to drive more responsive and interactive offerings. An IoT-based network able to manage this information is being investigated. Furthermore, real experiments to evaluate the subjective QoE of users are being performed thanks to real IoT devices (i.e., Arduino, Raspberry) fully programmable to develop applied research.