Mobile Instant Messaging (MIM) applications such as WhatsApp transformed human communication by enabling global exchange of various message types, such as text, image, video, or voice, at any time. Network providers are facing a substantial user base and network load which is especially high in group chats where each message needs to be distributed to each member. Due to end-to-end encryption, network operators must obtain knowledge about the communication and the resulting load on the network by other means, which makes it necessary to model the network traffic of MIM. In this work, we therefore present an approach to source traffic modeling for MIM. After identifying the building blocks of a Source Traffic Model (STM) for MIM, we address existing gaps through studies on MIM communication networks, contact proximity, media compression and payload size, as well as media file size distribution. Combining existing literature and our work, we present and implement a modular STM approach which can be used for developing STMs for MIM. Using an exemplary STM, we evaluate the daily network traffic per user. With this, we provide a comprehensive description of MIM in the network researching context and enable consideration of MIM in future network design.

Global connectivity demands extensive cellular network coverage, enabling instant data transmission to users worldwide. The growing need for faster, low-latency networks drives the global 5G rollout, aiming to cover one-third of the population by 2025. While 5G promises enhanced Quality of Service, network providers must balance customer expectations with political pressure and the costs of expansion. However, it remains unclear whether current 5G deployments adequately address connectivity challenges and deliver consistently high Quality of Experience. To investigate this, we analyze mobile network quality across Bavaria using over 225 million measurements from Opensignal. By correlating key performance metrics like throughput, latency, jitter, and packet loss with network generation and provider, we evaluate technology and service impacts on network quality. These findings are mapped to Bavaria's population to explore urbanization effects on performance and identify potential digital divides between urban and rural areas.

This paper explores the relationship between loading times and Quality of Experience (QoE) in Mobile Instant Messaging (MIM) applications. Using a web application that mimics MIM interfaces, we conducted a QoE study in which participants engaged with a virtual chat partner. We controlled image loading times during chatting and evaluated their impact on QoE, annoyance, and acceptance ratings. Although the results show no difference in the QoE ratings, they clearly show that longer delays lead to greater annoyance and lower user acceptance. These findings underscore the importance for MIM app providers to minimize loading times in order to increase user satisfaction and retention.

Data-driven QoE modelling using Machine Learning (ML) allows to reduce the modelling bias and to continuously integrate new QoE results into the QoE model, which can improve its generalizability. The downside is that the majority of ML models are black-box models, which prevent to obtain insights about QoE influence factors and their fundamental relationships that are highly relevant for researchers and providers of services and networks. However, recent advances in the field of eX-plainable Artificial Intelligence (XAI) resolve these issues. Thus, XAI allows to benefit from data-driven QoE modelling to obtain generalizable QoE models, and at the same time to understand what QoE factors are relevant and how they affect the QoE score. In this work, we showcase the feasibility of explainable data-driven QoE modelling for video streaming, since video streaming QoE has been well researched, and thus, allows us to validate our results. Finally, we discuss opportunities and challenges of deploying XAI for QoE modelling.

Streaming has become the dominant source of media consumption, which not only applies to the widely researched field of video streaming, but also to music streaming. Here, previous studies so far have only researched the impact of streaming aspects, such as stalling events or initial loading times, on the QoE of music streaming. However, when using a music streaming application, users are already facing waiting times along the click path before they can start the actual streaming. These waiting times are caused by browsing delays, e.g., during searching for songs or scrolling through playlists, and can potentially deteriorate the QoE of the music streaming application. In this work, we conduct an online QoE study to quantify the impact of these browsing delays with the support of an emulated mobile music streaming web app. We found that browsing delays have no impact on the music streaming QoE, which shows that users are able to clearly distinguish between the two main functionalities of such apps, namely, browsing and streaming. However, browsing delays significantly reduce the QoE of the entire music streaming application, to a similar extent as if QoE degradations happen during the actual streaming. This shows that both browsing and streaming are equally important and have to be considered when designing music streaming annlications.

Crowdsourced measurements solve the problem of being able to assess the performance of a communication network from an end-user perspective, but the new characteristics of the data pose new challenges for QoE modeling. In contrast to existing laboratory or network measurements, this type of measurement at the end user device primarily involves taking a large number of short sample measurements, which, however, are rich in measured parameters, including many user-, application-, and device-related parameters. To test the applicability and to facilitate the integration of such data, we applied four QoE models from the literature to 290k worldwide video streaming measurements from a commercial data set from August to October 2020. In this work, we will therefore first describe the crowdsourcing video streaming data set to provide insights into the properties of video streaming KPIs in the real world. Second, we run four popular QoE models using this data set, compare the resulting QoE scores, and derive the impact of individual KPIs for each model. We show that the models assess the QoE at least differently, but sometimes with contradicting statements. Reading this paper, it becomes evident that more work and subjective studies, based on real-world data like the one we have shown, are needed to extend the current QoE models.

Streaming made a lasting effect on the way our society consumes media in the last decade. While due to streaming the way we listen to music and podcasts has changed drastically, there are very few studies about its Quality of Experience (QoE) and possible influence factors. From video QoE studies, we know that, for example, undesirable stops of the stream (stalling events) have a significant impact on QoE. However, the way in which music and video streaming is consumed differs significantly, as music is often played in the background, and thus, the influence of stalling could be significantly different. Thus, this work evaluates the impact of stalling on music streaming QoE. Therefore, we conduct two crowdsourced user studies: In the first study, users have to rate four songs with different stalling patterns and evaluate the degree of impairments. Afterwards, we compare the ratings to the results of a lab study and show that they are highly correlated, and that crowdsourcing is a suitable way of measuring music streaming QoE. In addition, we conduct a second crowdsourcing study to investigate the influence of the user's attentiveness on QoE. Here, participants have to listen to one song with two stalling events, while one half of them had to transcribe a handwritten text with music playing in the background. The attentiveness shows no influence on the perceived streaming quality, but it shows a significant influence on the perceived quality degradation due to stalling events. Furthermore, considerably more stalling events were missed for workers who focused on the transcription. These results are an important step towards establishing new methods for investigating QoE in multimedia.

Evaluating the QoE of video streaming and its influence factors has become paramount for streaming providers, as they want to maintain high satisfaction for their customers. In this context, crowdsourced user studies became a valuable tool to evaluate different factors which can affect the perceived user experience on a large scale.In general, we observed that most of these crowdsourcing studies either use an in vivo or an in vitro design. In vivo design means that the study participant has to rate the QoE of a video that is embedded in an application similar to a real streaming service, e.g., YouTube or Netflix. In vitro design refers to a setting, in which the video stream is separated from a specific service and thus, the video plays on a plain background. Although these designs vary widely, the results are often compared and generalized.Therefore, in this work, we investigate the influence of these two study design alternatives on the perceived QoE. In crowdsourced user studies, participants rate the video streaming with respect to different stalling patterns (no stalling, different positions) and study designs (in vivo or in vitro). Contrary to our expectations, the results indicate that there is statistically no significant influence of the study design on the perceived video QoE and acceptance. In addition, we found that the in vivo design does not reduce the test takers' attentiveness.

The number of smartphones connected to wireless networks and the volume of wireless network traffic generated by such devices have dramatically increased in the last few years, making it more challenging to tackle wireless network monitoring applications. The high-dimensionality of network data provided by current smartphone devices opens the door to the massive application of machine learning approaches to improve different wireless networking applications. In this paper we study the specific problem of Quality of Experience (QoE) prediction for popular smartphone apps, using machine learning models and in-smartphone measurements. We evaluate and compare different models for the analysis of smartphone generated data, including single models as well as machine learning ensembles such as bagging, boosting and stacking. Results suggest that, while decision-tree based models are the most accurate single models to predict QoE, ensemble learning models, and in particular stacking ones, are capable to significantly increase accuracy prediction and overall classification performance.

Internet Service Providers need a thorough understanding of a service to maximize the Quality of Experience (QoE) of their customers by network management. Instead of quantifying the user satisfaction with long and cost-intensive subjective user studies, the QoE can often be estimated with the help of dedicated measurements of application and network parameters. We designed a QoE measurement tool for the popular audio streaming service Spotify that runs inside a Docker software container. The container is able to run headlessly as active measurement probe and emulates a user who is streaming audio files via Spotify. While streaming, network and application parameters are collected that have a high correlation to the user's QoE. The results of the measurements are used to characterize audio streaming in Spotify on application and network layer, and to evaluate important QoE factors.

Monitoring the Quality of Experience (QoE) undergone by cellular network customers has become paramount for cellular ISPs, who need to ensure high quality levels to limit customer churn due to quality dissatisfaction. This paper tackles the problem of QoE monitoring, assessment and prediction in cellular networks, relying on end-user device (i.e., smart-phone) QoS passive traffic measurements and QoE crowdsourced feedback. We conceive different QoE assessment models based on supervised machine learning techniques, which are capable to predict the QoE experienced by the end user of popular smartphone apps (e.g., YouTube and Facebook), using as input the passive in-device measurements. Using a rich QoE dataset derived from field trials in operational cellular networks, we benchmark the performance of multiple machine learning based predictors, and construct a decision-tree based model which is capable to predict the per-user overall experience and service acceptability with a success rate of 91% and 98% respectively To the best of our knowledge, this is the first paper using end-user, in-device passive measurements and machine learning models to predict the QoE of smartphone users in operational cellular networks.