Cloud gaming (CG) enables on-demand, affordable, anytime, anywhere access to games streamed over heterogeneous access networks (HANs). With CG, AAA games can be played on inexpensive devices such as smartphones at home, during commutes, or in public spaces via HANs like Wi-Fi, 4G, and 5G. While this broad service accessibility is a key advantage for CG, HANs are often affected by unpredictable network conditions, including congestion and handovers, which can impair Quality of Service (QoS) metrics such as round-trip time (RTT), packet loss (PL), and jitter. Such QoS degradation can immediately reduce audiovisual quality and responsiveness, interrupting active play and resulting in a poor user experience. This thesis examines how QoS factors affect users’ Quality of Experience (QoE) for CG services provisioned over HANs. QoE, a user-centered metric, reflects how users perceive the quality of an application or service and is typically measured using objective models developed from extensive subjective test data. In the context of CG, QoE helps to understand how degradations in gaming quality due to QoS issues are reflected in users’ experience. The thesis specifically addresses two cases: mobile cloud gaming (MCG) and virtual reality cloud gaming (VRCG). MCG involves smartphone-based, on-the-go gaming over HANs, which are particularly susceptible to mobility-related impairments such as fluctuating network quality. In contrast, VRCG centers on immersive, embodied interaction through head-mounted displays (HMDs) that require responsiveness, frame stability, and input–output synchronization for seamless play. To address these challenges, this thesis proposes, develops, and validates novel methods for QoE measurement and prediction for MCG and VRCG under HANs conditions. First, we conducted extensive subjective tests with 116 unique participants, covering 37 and 28 realistic network conditions for MCG and VRCG, respectively. Controlled laboratory environments were established to emulate network degradations due to RTT, PL, and jitter, ensuring standardized, repeatable experimentation. Complementing this, the thesis develops and validates ALTRUIST, a multi-platform orchestration tool that automates test execution, questionnaire control, traffic capture, data labeling, and experimental condition management across heterogeneous devices. These testbeds, together with ALTRUIST, enabled 4 comprehensive subjective studies and yielded 4 novel datasets on MCG and VRCG. Second, this thesis also proposes, develops, and validates novel objective QoE models for MCG and VRCG. These models account for the aforementioned QoS factors using datasets from our subjective tests. For MCG, we developed three regression models: linear, polynomial, and spline. Each used two datasets. Our results show that QoE can be predicted with high accuracy in all cases. In particular, the spline regression model achieved a mean absolute error (MAE) of 0.19. For VRCG, three regression models were developed and validated using a combined dataset from two game genres: shooter and casual. Our results show that QoE can be accurately predicted from QoS factors. For example, the non-linear regression model achieved an average MAE of 0.22. Our QoE models offer practical methods for QoE prediction for MCG and VRCG over HANs. Finally, this thesis presents novel Bayesian analysis results. These results investigate probabilistic and complex relationships among several factors, including interactivity, video quality, audio quality, and overall QoE. The Bayesian analysis provides further insight into how network degradation affects the gaming experience in both MCG and VRCG over HANs.

The outcomes of this thesis resulted in 10 peer-reviewed publications, a patent, and contributions to two international standards. These standards were published by the International Telecommunications Union. 

Cloud-based social eXtended Reality (XR) services are the cornerstone for realizing the promises of the Metaverse. These services hosted either on datacenters or edge, will demand stringent mobile network quality of service (QoS) to operate effectively and provide an acceptable user quality. It becomes fundamental to study how mobile networks QoS factors round-trip time (RTT), packet loss (PL), and jitter affect these services by measuring their effect on users' perceived quality of experience (QoE). Subjective QoE assessment involves carefully controlled laboratory environments to generate the desired conditions between a large set of users. The requirements for a cloud-based social XR service lab-setup are complex: Identify a reliable streaming service, a customizable VR application, emulate network conditions, define activities or tasks for users to perform, collect their data, label it; all while mitigating possible human mistakes. To address these requirements, we present an effective technical setup that can consistently repeat the same conditions between users and that can be easily replicated to other labs conducting cloud-based social XR research.

Cloud gaming (CG) services are a key enabler of more accessible and affordable gaming experiences for users worldwide. However, to deliver these services with sufficient Quality of Experience (QoE) over heterogeneous access networks, stakeholders, including network and cloud service providers, and game developers, must first understand their specific performance requirements. For that, this study investigates QoE of PC-based CG under controlled and repeatable network conditions, emulating a range of impairments including round-trip time (RTT), packet loss (PL), random jitter (RJ), bursty jitter (BJ), and bitrate. Two representative games, Terraria and CounterStrike 2, were tested across 15 network conditions, streamed using the Moonlight/Sunshine platform. Thirty participants rated their QoE, with orchestration and data collection managed by the ALTRUIST tool. Results show that low bitrate (0.5Mb/s), high RTT(402ms), bursty jitter (702ms), and high RJ (std=3) significantly degrade QoE, especially in QoS-sensitive games. Additionally, we evaluate mobile-CG trained QoE models on PC-CG data and observe significant prediction errors, highlighting the importance of accounting for differences in streaming platform and context when estimating QoE. These findings provide valuable insights for stakeholders optimizing PC-CG over HANs.

Cloud gaming (CG) services are a key enabler of more accessible and affordable gaming experiences for users worldwide. However, to deliver these services with sufficient Quality of Experience (QoE) over heterogeneous access networks, stakeholders, including network and cloud service providers, and game developers, must first understand their specific performance requirements. For that, this study investigates QoE of PC-based CG under controlled and repeatable network conditions, emulating a range of impairments including round-trip time (RTT), packet loss (PL), random jitter (RJ), bursty jitter (BJ), and bitrate. Two representative games, Terraria and Counter-Strike 2, were tested across 15 network conditions, streamed using the Moonlight/Sunshine platform. Thirty participants rated their QoE, with orchestration and data collection managed by the ALTRUIST tool. Results show that low bitrate (0.5Mb/s), high RTT (402ms), bursty jitter (702ms), and high RJ (std=3) significantly degrade QoE, especially in QoS-sensitive games. Additionally, we evaluate mobile-CG trained QoE models on PC-CG data and observe significant prediction errors, highlighting the importance of accounting for differences in streaming platform and context when estimating QoE. These findings provide valuable insights for stakeholders optimizing PC-CG over HANs.

In the future, extended reality (XR) applications will be hosted on cloud and edge infrastructures and streamed over heterogeneous access networks such as Wi-Fi and 6G. These infrastructures promise ubiquity but are prone to stochastic conditions, such as network congestion and wireless signal fading and attenuation, that can be detrimental to the quality of experience (QoE) regarding XR applications. This paper presentsextensive and novel results assessing the impact of networkconditions (N = 20) on users’ QoE via realistic subjective tests(N = 28) involving factors such as round-trip time (RTT), jitter(RJ), and packet loss (PL) in a social XR application context. Our results indicate that social XR applications require stringent quality of service to support users’ QoE. We demonstrate that RTT values up to 77 ms do not significantly impact users’ QoE. Furthermore, combined (PL and RTT) values lead to significant QoE degradation when RTT values exceed 77 ms and PL exceeds 2%. We also demonstrate that even minimal jitter values, with 1 standard deviation beyond 52 ms RTT values,can lead to significant QoE degradations. Furthermore, jitter values exceeding 3 standard deviations for 27ms RTT value and beyond should be avoided. Finally, based on network traffic data between Sweden and various AWS data centers in Europe, we show that social XR applications can be hosted at a few datacenter locations with minimal QoE impact for wired network connections. However, due to high jitter values, both 4G and 5G networks are not conducive to users’ QoE.

Virtual reality cloud-based gaming (VRCG) services are becoming widely available on virtual reality (VR) devices delivered over computer networks.VRCG brings users worldwide an extensive catalog of games to play anywhere and anytime. Delivering these gaming services in existing broadband mobile networks is challenging due to their stochastic nature and the user perceived Quality of Experience (QoE)' sensitivity towards them. More research is needed regarding developing effective methods to measure the impact of network QoS factors on users' QoE in the VRCG context. Therefore, this paper proposes, develops, and validates three novel regression models trained on a real dataset collected via subjective tests (N=30); the dataset contains subjective users' QoE ratings regarding VR shooters affected by network conditions (N=28), such as round-trip time (RTT), random jitter (RJ), and packet loss (PL). Our findings reveal that due to the nonlinear relationship of (RTT and RJ) tested together, nonlinear(mean absolute error (MAE)=0.14) and polynomial (MAE=0.15) regression models have the best performance; yet, simple linear regression model(MAE=0.19) is also suitable to predict QoE for VRCG. Further, we found that the model's most important feature is RTT, followed by (RTT, RJ). Finally, our models' prediction of QoE for real-world traffic measurements suggests that mobile network traffic (4G, 5G non-standalone, 5G standalone) provides a 2.5 \(\leq MOS\_{QoE} \leq\) 3.0 experience for VRCG, while 4.2 \(\leqMOS\_{QoE} \leq\) 4.4  for wired connections, suggesting the need for improvements in the current commercial 5G network deployments to deliverVRCG.