Haptic teleoperation systems mark a critical breakthrough in remote manipulation technologies, delivering immersive user experiences through precise control and tactile feedback. This paper aims to explore the interplay between Quality of Service (QoS) network metrics, such as latency, jitter, and packet loss, and Quality of Experience (QoE) features, including immersion, control, and engagement. This investigation is conducted through a combination of objective measurements and subjective evaluations over a private Fifth Generation Standalone (5G SA) network. In addition, it aims to realize the extent to which transport-layer protocols operating over IP, such as TCP and UDP, as well as environmental factors like indoor and outdoor settings, influence system performance.Results confirm the well-established trade-off between reliability and latency in transport protocols, with TCP offering higher reliability in controlled indoor environments, and UDP exhibiting better responsiveness in dynamic outdoor scenarios. While these findings align with existing knowledge, their empirical validation in the context of immersive haptic applications over a private 5G SA network reinforces their relevance. In addition, QoE metrics were found to be linked to QoS indicators, highlighting the importance of balancing speed, stability, and reliability. These findings provide valuable insights towards designing adaptive teleoperation systems capable of dynamically optimizing performance under diverse conditions.

In this paper, we investigate the effects of asynchrony between the visual and haptic feedback in virtual reality (VR) on user experience, specifically focusing on understanding users' awareness of this asynchrony and its effect on their level of satisfaction. Using Unreal Engine, we created an experimental setup to adjust the timing between these sensory inputs. Our experiment featured a VR dodge game that provides haptic feedback on the body when the player is hit by a multitude of virtual objects. Conducting a targeted, small-scale user study, we aim to understand in what ways an introduced asynchrony influences the VR experience. The results highlight the perceptibility of asynchrony, which significantly affects the overall user experience. Nonetheless, we also find an asymmetry that benefits scenarios where haptic feedback precedes visual cues. Furthermore, our findings suggest that users can generally accept minor levels of asynchrony without significant disadvantages to their satisfaction. However, it is interesting to note that even when users cannot explicitly identify any asynchrony, they might still experience a slight decrease in satisfaction.

Fifth Generation (5G) networks are becoming the norm in the global telecommunications industry, and Mobile Network Operators (MNOs) are currently deploying 5G alongside their existing Fourth Generation (4G) networks. In this paper, we present results and insights from our large-scale measurement study on commercial 5G Non Standalone (NSA) deployments in a European country. We leverage the collected dataset, which covers two MNOs in Rome, Italy, to study network deployment and radio coverage aspects, and explore the performance of two use cases related to enhanced Mobile Broadband (eMBB) and Ultra-Reliable Low Latency Communication (URLLC). We further leverage a machine learning (ML)-based approach to model the Dual Connectivity (DC) feature enabled by 5G NSA. Our data-driven analysis shows that 5G NSA can provide higher downlink throughput and slightly lower latency compared to 4G. However, performance is influenced by several factors, including propagation conditions, system configurations, and handovers, ultimately highlighting the need for further system optimization. Moreover, by casting the DC modeling problem into a classification problem, we compare four supervised ML algorithms and show that a high model accuracy (up to 99%) can be achieved, in particular, when several radio coverage indicators from both access networks are used as input. Finally, we conduct analyses towards aiding the explainability of the ML models.

In the context of massive machine-type communications (mMTCs), the narrowband Internet-of-Things (NB-IoT) technology is envisioned to efficiently and reliably deal with massive device connectivity. Hence, it relies on a tailored random access (RA) procedure, for which theoretical and empirical analyses are needed for a better understanding and further improvements. This article presents the first data-driven analysis of NB-IoT RA, exploiting a large-scale measurement campaign. We show how the RA procedure and performance are affected by network deployment, radio coverage, and operators' configurations, thus complementing simulation-based investigations, mostly focused on massive connectivity aspects. A comparison with the performance requirements reveals the need for procedure enhancements. Hence, we propose a machine learning (ML) approach and show that RA outcomes are predictable with good accuracy by observing radio conditions. We embed the outcome prediction in an RA-enhanced scheme and show that optimized configurations enable power consumption reduction of at least 50%. We also make our data set available for further exploration, toward the discovery of new insights and research perspectives.

In recent years, Downlink (DL) throughput estimation in Mobile Broadband (MBB) networks has gained immense popularity and it is expected to become a vital component of the upcoming fifth generation (5G) systems. Plentiful adaptive video streaming algorithms greatly rely on accurate DL throughput predictions to adapt their mechanisms and ensure high Quality of Service (QoS) to the end-users. Thus far, conventional DL throughput estimation approaches, also known as speed tests, require an extensive exchange of TCP traffic over the network for an allocated time duration. While such tools appear to deliver trustworthy results, they turn out to be inefficient when mobile subscriptions with limited data plans are engaged. In this paper, we propose a supervised Machine Learning (ML) solution for DL throughput estimation that aims at delivering highly accurate predictions while significantly limiting the over-the-air data consumption. We capture the network performance metrics by exploring both crowdsourced and controlled testing methodologies. We leverage RTR-NetTest, a platform of broadband measurements provided by the Austrian Regulatory Authority for Broadcasting and Telecommunications (RTR), and MONROE-NetTest, its counterpart wrapper built as an Experiment as a Service (EaaS) on top of Measuring Mobile Broadband Networks in Europe (MONROE). Results reveal that our solution can achieve a 39.7% reduction in terms of data consumption while delivering a Median Absolute Percentage Error (MdAPE) of 5.55%. We further show that accuracy can be traded-off, for example, a significant data consumption reduction of 95.15% can be achieved for a MdAPE of 20%.

Hyperparameter optimization is an important but often ignored part of successfully training Neural Networks (NN) since it is time consuming and rather complex. In this paper, we present HINDSIGHT, an open-source framework for designing and implementing NN that supports hyperparameter optimization. HINDSIGHT is built entirely in R and the current version focuses on Long Short Term Memory (LSTM) networks, a special kind of Recurrent Neural Networks (RNN). HINDSIGHT is designed in a way that it can easily be expanded to other types of Deep Learning (DL) algorithms such as Convolutional Neural Networks (CNN) or feed-forward Deep Neural Networks (DNN). The main goal of HINDSIGHT is to provide a simple and quick interface to get started with LSTM networks and hyperparameter optimization.

Crowdsourcing mobile network performance evaluation is rapidly gaining popularity, with new applications aiming to deliver more accurate and reliable results every day. From the perspective of end-users, these utilities help them estimate the performance of their service provider in terms of throughput, latency and other key performance indicators of the network. In this paper, we build ORCA: Operator Classifier, a Machine Learning (ML) based framework to define and determine the behavior of Mobile Network Operators (MNOs) from crowdsourced datasets. We investigate whether one can differentiate MNOs by using crowdsourced end-to-end network measurements. We consider different performance metrics (e.g. Download (DL)/Upload (UL) data rate, latency, signal strength) and study the impact of them individually but also collectively on differentiating MNOs. We use RTR Open Data, an open dataset of broadband measurements provided by the Austrian Regulatory Authority for Broadcasting and Telecommunications (RTR), to characterize the three major mobile native operators and two virtual operators in Austria. Our results show that ORCA can be used to identify patterns between various mobile systems and disclose their differences from the end-user perspective.