The convergence of the Internet of Things (IoT) and the metaverse creates systems where a Digital Twin (DT) sits between physical and virtual worlds. Keeping them in sync is hard when actions arrive at the same time with different priorities and delays. We present a DT template that supports two-way, asynchronous interactions and a stack of three policies: confirm matching intents, resolve conflicts with context (domain weights and timeliness), and rollback on invariant violations. We validate the approach in a smart building with DTs representing doors, windows, and lights operating within policy domains that include security, comfort, and energy efficiency. Without policies, conflicting windows already reach 33-56% with only two users for different request rates. With the policy stack, the share of conflicts resolved by policy increases with the reliability gap and can approach all the conflicts, reducing rollbacks. Finally, under asymmetric placement, by simulating the DT at different points in the network (e.g., edge close to physical devices and cloud close to the metaverse), rollbacks shift to the slower side but leave the overall resolution essentially unchanged.

It is well-known that, without access to application-layer parameters controlled by Over-The-Top (OTT) providers, Mobile Network Operators (MNOs) struggle to accurately predict customers’ Quality of Experience (QoE). While some previous proposals have suggested interaction between OTTs and MNOs, they have faced challenges in terms of practical implementation and limited application scenarios. This work aims to advance these solutions with two key contributions. First, following the Open Radio Access Network (O-RAN) architecture, we propose adding components that integrate a machine learning (ML)-based QoE prediction model, deployed by the MNO, into the O-RAN system. By establishing specific data-sharing interfaces between OTTs and MNOs, our approach helps MNOs overcome the limitations in updating their quality prediction modules. Second, we present a network-aware, ML-driven QoE prediction model that captures the relationship between the resulting QoE and various network parameters, such as signal-to-interference-noise ratio (SINR), channel quality indicator (CQI), network resource blocks (RBs), throughput, and device mobility. Among seven considered ML regressors, the Gradient Boosting (GB) achieved the highest QoE prediction performance in terms of R 2 (0.906) and RMSE (0.259).

Human Digital Twins (HDTs) are increasingly adopted across various domains, yet their application to network and service management remains limited. Nevertheless, HDTs offer significant potential for optimizing service configurations based on human behavior, preferences, and profiles. In this paper, we analyze the role of HDTs in network and service management, identifying key functionalities such as collaborative learning for user modelling, Quality of Experience (QoE) and emotion prediction, application personalization, and behavioral forecasting for network “what-if” analysis. We propose an architectural framework designed to monitor user status and generate a corresponding digital replica that interacts with other network components to enhance service delivery. Our solution integrates collaborative learning for QoE modelling and applies it to service optimization. By aggregating user data from multiple HDTs, the approach improves prediction accuracy and resource optimization. Extensive performance evaluations demonstrate that the proposed collaborative HDT framework enhances the final utility function that considers perceived quality and resource usage by 27% compared to non-collaborative methods.

This paper presents the results of a subjective assessment investigating the impact of multiple factors (network, avatar, and player role) on the perceived Quality of Experience (QoE) in a multi-user collaborative virtual reality (VR) game. Forty test participants collaborated in pairs to complete a cooking VR game, wearing a Meta Quest Pro headset, under variable network conditions (no impairments or delayed network traffic), using diverse types of avatars (cartoon-style and humanoid), and interpreting different roles (teacher and student). A humanoid custom avatar has been implemented that replicates the user’s facial expressions and body movements to investigate whether the introduction of non-verbal emotional communication within a VR environment influences the perceived user experience. Quality-and emotion-related subjective metrics were rated by test participants at the end of each test session, and the computed results show that the cartoon-like avatar, being lightweight, provides the highest perceived QoE even when the network was impaired. On the other hand, while the QoE using the humanoid avatar lowers with the introduction of network distortions (because of the larger amount of data required to replicate the facial expressions), the ability to see the facial expressions of the partner prevents a greater reduction of user experience due to the network issues.

Accurate models are necessary for the continuous estimation of the Quality of Experience (QoE), which is crucial for delivering successful multimedia services to end-users. These models are developed from subjective test data, which very often provide only specific aspects of the experience, i.e., a partial view (PV). Each PV conveys a relationship between a specific set of influence factors and the perceived QoE, limiting the applicability of the derived model to other application scenarios not considered in the initial subjective tests. To extend the applicability of the developed models, this paper introduces a multi-view (MV) learning framework that enhances QoE prediction by integrating complementary information from multiple perspectives obtained from different subjective tests. We leverage a fully connected deep neural network with two initially independent branches and an intermediate fusion layer to combine insights from separate feature sets, improving predictive accuracy while preserving data privacy. Our model is trained on a synthetic data set derived from the TID2008 image database, ensuring a controlled yet representative evaluation environment. On the one hand, the results demonstrate that the MV technique outperforms all PV configurations. On the other hand, the MV approach achieves QoE estimation performance comparable to the single-view (SV) model, in which one single branch analyzes the full set of impact factors. In particular, the largest performance gain (6.15% – 142.69%) across most evaluation metrics occurred when the input data set is equally divided between the two separate views.

Point cloud quality assessment (PCQA) is a critical research area focused on evaluating the perceptual Quality of Experience (QoE) of point clouds to enhance visual experiences of immersive multimedia applications for end users. To prevent the complex computations on 3D data applied by model-based methods, projection-based models have been developed to estimate the QoE by analysing 2D projection views of the point cloud. In this paper, we propose a novel projection-based No-Reference (NR) PCQA method, called Multi-View Adaptive Weighting Point Cloud Quality Assessment (MVAW-PCQA), to predict the QoE of distorted point clouds using six 2D projection views as the input of a convolutional neural network (CNN) architecture. First, multi-view involves independently extracting features from multiple projection views of a point cloud, guaranteeing view-specific features are learned without prematurely mixing spatial information, and preserving the unique contributions of each projection view to the final quality prediction. Then, an adaptive weighting fusion mechanism combines the features extracted from the different projection views by learning their relative importance. This design enables the model to focus on the most informative projections for predicting the point cloud quality. The experimental results demonstrate that our method outperforms state-of-the-art NR-PCQA methods on the SJTU-PCQA dataset in terms of root mean square error (RMSE) and correlation coefficients (Pearson, Spearman, and Kendall), while adopting a lightweight design with a reasonable number of parameters for the trained neural network.

White blood cell (WBC) subtype classification is a critical step in monitoring an individual’s health. However, it remains a challenging task due to the significant morphological variability of WBCs and the domain shift introduced by differing acquisition protocols across hospitals. Numerous approaches have been proposed to mitigate domain shift, including supervised and unsupervised domain adaptation, as well as domain generalisation. These methods, however, require a suitable amount of representative target images, even if unlabelled, or a suitable amount of images from multiple sources, which may not be feasible due to privacy regulations. In this study, we explore an alternative paradigm, known as Distributed Collaborative Machine Learning (DCML), which consists of exploiting images from different sources in a privacy-preserving setup. Although DCML methods seem well suited to this application, to the best of our knowledge, they have not been used for this task or to address the above-mentioned issues. However, we argue that DCML deserves further consideration in medical images as a potential alternative solution against domain shift in a privacy-preserving setup. To substantiate our view, we consider three DCML methods: early and late fusion and federated learning approaches, each offering distinct trade-offs in terms of training constraints, computational overhead and communications costs. We then conduct an extensive, cross-dataset experimental evaluation on four benchmark datasets and provide evidence that even simple implementations of DCML methods can effectively mitigate domain shift in WBC classification tasks.

The increasing demand for quality from multimedia service users is very often addressed by adding more resources (bandwidth and processing power). However, not always does this approach bring an improvement in the perceived quality, whereas it frequently implies an increase in energy consumption (and subsequent higher greenhouse gas emissions). Accordingly, in this paper, we propose a solution to dynamically allocate resources in a 5G heterogeneous network scenario, which aims to identify a trade-off between the overall QoE perceived by the users served by the network when consuming video content and the overall network energy consumption. We considered three types of devices (TV, laptop, and smartphone) for which appropriate QoE and energy consumption models are defined. Extensive simulations have been performed by assigning different levels of importance to QoE and energy. The achieved results show that the network energy consumption can be more than halved by keeping satisfactory QoE. This is particularly true for smartphone users, whereas TV and laptop users have the freedom to choose based on their sensitivity towards sustainability.

Multi-view (MV) learning is a machine learning technique for improving generalization efficiency by learning from different feature subsets derived from multiple sources. We believe this approach can help in Quality of Experience (QoE) modelling by integrating knowledge from different datasets generated by subjective tests conducted for the same or similar applications considering different QoE Influence Factors (IFs). To investigate this subject, in this paper, we present the experiments conducted starting from a complete dataset related to Web browsing sessions that has been artificially divided into two distinct subsets (views). The proposed MV learning approach implements a data fusion technique to integrate extracted features from different views into a unified feature space. To achieve a complete experiment on the entire problem space, all possible combinations of IFs (features) in two distinct partial views (PVs) are considered and trained in the MV approach; the full view (FV) approach, which utilizes the complete dataset, is also considered for performance comparison. Experimental results show the QoE estimation performance achieved by the MV (0.69) is comparable with that of the FV (0.72), although the 2 single views were used for training in the MV case. Moreover, the performance enhancement achieved by the MV compared with the PV is most noticeable when a lower number of features is used to train the models.

This paper proposes NetRate, an Android system monitoring application that can measure and collect network and energy information from Android devices and enables end users to rate the Quality of Experience (QoE) of selected applications. Moreover, NetRate includes a VPN service to limit the network capabilities of the end device to simulate constrained network conditions. NetRate supports the conduction of QoE research studies in the laboratory and the field by collecting different kinds of data from the end device, assisting in setting test conditions (e.g., with the VPN), and collecting subjective feedback.

The utilization of user’s facial- and speech-related features for the estimation of the Quality of Experience (QoE) of multimedia services is still underinvestigated despite its potential. Currently, only the use of either facial or speech features individually has been proposed, and relevant limited experiments have been performed. To advance in this respect, in this study, we focused on WebRTC-based videoconferencing, where it is often possible to capture both the facial expressions and vocal speech characteristics of the users. First, we performed thorough statistical analysis to identify the most significant facial- and speech-related features for QoE estimation, which we extracted from the participants’ audio-video data collected during a subjective assessment. Second, we trained individual QoE estimation machine learning-based models on the separated facial and speech datasets. Finally, we employed data fusion techniques to combine the facial and speech datasets into a single dataset to enhance the QoE estimation performance due to the integrated knowledge provided by the fusion of facial and speech features. The obtained results demonstrate that the data fusion technique based on the Improved Centered Kernel Alignment (ICKA) allows for reaching a mean QoE estimation accuracy of 0.93, whereas the values of 0.78 and 0.86 are reached when using only facial or speech features, respectively.

In this article, we propose a Hand Gesture Recognition (HGR) system based on a novel deep transformer (DT) neural network for media player control. The extracted hand skeleton features are processed by separate transformers for each finger in isolation to better identify the finger characteristics to drive the following classification. The achieved HGR accuracy (0.853) outperforms state-of-the-art HGR approaches when tested on the popular NVIDIA dataset. Moreover, we conducted a subjective assessment involving 30 people to evaluate the Quality of Experience (QoE) provided by the proposed DT-HGR for controlling a media player application compared with two traditional input devices, i.e., mouse and keyboard. The assessment participants were asked to evaluate objective (accuracy) and subjective (physical fatigue, usability, pragmatic quality, and hedonic quality) measurements. We found that (i) the accuracy of DT-HGR is very high (91.67%), only slightly lower than that of traditional alternative interaction modalities; and that (ii) the perceived quality for DT-HGR in terms of satisfaction, comfort, and interactivity is very high, with an average Mean Opinion Score (MOS) value as high as 4.4, whereas the alternative approaches did not reach 3.8, which encourages a more pervasive adoption of the natural gesture interaction.

The increasing use of ICT has raised concerns about its negative impact on energy consumption and$CO_{2}$emissions. To address this issue, there is a need to better understand the trade-off between Quality of Experience (QoE) and sustainable video streaming services. In this study, we designed and conducted a subjective assessment to investigate the impact of video resolution, different types of luminance, and different end devices on the QoE and energy consumption of video streaming services. Then, we applied statistical models (Analysis of Variance and t-test) to subjective data to find out what factors influence the QoE the most and consume more energy. The obtained results suggest that under specific conditions (e.g., dark or bright ambient, low device backlight luminance, small-screen device) the users could be encouraged towards a trade-off between acceptable QoE and sustainable (green) choices because spending more energy (e.g., streaming higher-quality video) would not provide noticeable QoE enhancement.

WebRTC-based applications allow for real-time communications that are subject to network impairments affecting the end user's Quality of Experience (QoE). In this paper, we conducted subjective tests involving 20 people to investigate the conversational quality of a two-party WebRTC-based audiovisual telemeeting service. A dedicated system was implemented to introduce controlled network impairments (delay, jitter, and packet loss) to impair the communication between the parties. In addition, test participants had to rate the perceived QoE for the audio, the video, and the overall service, as well as the three emotional dimensions, i.e., valence, arousal, and dominance. Extensive results were obtained regarding the impact of the network impairments on the multimedia quality, the emotional dimensions, and the communication feasibility.

The Metaverse provides a novel experience to the user, by opening the doors to social-based multiuser environments merging physical reality with digital virtuality. In this paper, we present an initial analysis of the Quality of Experience (QoE) in the Metaverse. We first consider traditional influence factors (human, system, and context). Then, we introduce the social and economic dimensions of the Metaverse as additional factors to be considered for QoE assessment. Finally, we discuss what QoE assessment methods can be more suitable for Metaverse applications, with a particular focus on implicit assessment methods (e.g., physiological, human cognitive, affective behaviour).

This article investigates the possibility to estimate the perceived Quality of Experience (QoE) automatically and unobtrusively by analyzing the face of the consumer of video streaming services, from which facial expression and gaze direction are extracted. If effective, this would be a valuable tool for the monitoring of personal QoE during video streaming services without asking the user to provide feedback, with great advantages for service management. Additionally, this would eliminate the bias of subjective tests and would avoid bothering the viewers with questions to collect opinions and feedback. The performed analysis relies on two different experiments: i) a crowdsourcing test, where the videos are subject to impairments caused by long initial delays and re-buffering events; ii) a laboratory test, where the videos are affected by blurring effects. The facial Action Units (AU) that represent the contractions of specific facial muscles together with the position of the eyes' pupils are extracted to identify the correlation between perceived quality and facial expressions. An SVM with a quadratic kernel and a k-NN classifier have been tested to predict the QoE from these features. These have also been combined with measured application-level parameters to improve the quality prediction. From the performed experiments, it results that the best performance is obtained with the k-NN classifier by combining all the described features and after training it with both the datasets, with a prediction accuracy as high as 93.9% outperforming the state of the art achievements.

The physical context, i.e., the characteristics of location and space where the multimedia service is consumed, may strongly influence the overall perceived Quality of Experience (QoE). In this paper, we investigate the effects of ambient illumination and noise on two multimedia consumption scenarios: watching a video on TV and reading a comic strip on tablet. To this aim, we organized an experiment considering different combinations of ambient illumination and introducing a disturbing noise. Then, we conducted a subjective quality assessment involving 20 people, who were asked to rate the perceived QoE using the 5-level Absolute Category Rating (ACR) quality scale and to express their emotions completing the Self-Assessment Manikin (SAM) questionnaire. The impact of illumination and noise on ACR ratings and SAM scores is evaluated computing the Multivariate Analysis of Variance (MANOVA). Finally, a QoE prediction model based on illumination and noise context factors is presented.

As known from everyday contexts of multimedia usage, suddenly occurring quality impairments are capable of causing strong negative emotions in human users. This is particularly the case if the displayed content is highly relevant to current motives and behavioral goals. The present study investigated the effects of visual degradations on quality perception and emotional state of participants who were exposed to a series of short video clips. After each video playback, participants had to decide whether a certain event happened in the video. For data collection, subjective measures of quality and emotion were complemented by behavioral measures derived from capturing participants' spontaneous facial expressions. For data analysis, two general approaches were combined: First, a multivariate analysis of variance approach allowed to examine the effects of visual degradation factors on perceived quality and subjective emotional dimensions. It mainly revealed that perceived quality and emotional valence were both sensitive to degradation intensity, whereas the impact of degradation length was limited when task-relevant video content had already been obscured. Second, using a machine learning approach, an automatic Video Quality of Experience (VQoE) prediction system based on the recorded facial expressions was derived, demonstrating a strong correlation between facial expressions and perceived quality. Hereby, estimates of VQoE might be delivered in an objective, continuous and concealed manner, thus diminishing any further need for subjective self-reports.

This paper investigates the applicability of QoE management on Smart City services. First, we analyze how quality management is implemented for traditional public services (services for which the presence of ICT technologies is limited or not necessary). We then propose a potential framework for the QoE management in Smart City services (traditional public services supported by an important, but not essential, presence of ICT systems), which is based on ICT systems for QoE prediction and management. Finally, we highlight the challenges to be addressed in the near future.