Optimizing Emerging Game Streaming Applications for Mobile Platforms
Restricted (Penn State Only)
- Author:
- Bhuyan, Sandeepa
- Graduate Program:
- Computer Science and Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- February 22, 2024
- Committee Members:
- Chitaranjan Das, Program Head/Chair
Chitaranjan Das, Chair & Co-Dissertation Advisr
Mahmut Kandemir, Dissertation Co-Advisor
C Lee Giles, Outside Unit & Field Member
Mahanth Gowda, Major Field Member - Keywords:
- Game Streaming
Cloud Gaming
Video Game Processing
Mobile platforms
Performance
Energy-efficiency - Abstract:
- Over the past five decades, video games have evolved tremendously since their inception. Advances in computer graphics techniques and graphics hardware have elevated video games from a rudimentary dot on an oscilloscope to highly immersive and realistic experiences of the present day. The advent of 5G era and the recently emergent cloud gaming services (such as Microsoft xCloud, Sony PS Now, NVIDIA GeForce Now, etc.), encompassing a wide collection of games, which offer Games-as-a-Service (GaaS), have made video games more accessible to mobile users. Especially, cloud gaming services have offered the benefits of experiencing desktop/console quality games on mobile devices. However, streaming high-resolution high-fidelity games (4K UHD) within real-time constraints (60 FPS) often suffers from lower Quality of Service (QoS) and consumes more battery power on mobile devices. Due to their real-time and user-interactive nature, the prior traditional video streaming optimizations are not directly applicable to cloud gaming. Therefore, given the power-constraints and resource limitations of mobile SoCs, it is critical to identify and optimize the performance and energy inefficiencies of this emerging use case which can assist in designing the next generation mobile devices for providing acceptable QoS, while improving the energy efficiency. The target of this dissertation is to enhance the performance and energy efficiency of game streaming applications on mobile platforms. Towards this, the dissertation explores three intertwined tasks. (1) The first task involves systematically investigating the <QoS, Energy> design pairs on the end-to-end game streaming pipeline across the cloud, network, and mobile/edge devices to understand the individual contributions of the different pipeline stages for improving the overall QoS and energy efficiency. The insights from this task lead to designing energy-efficient and network-friendly solutions enhancing the frame encoding efficiency and reducing the game frame drops during network transmission for the next two tasks. (2) The second task involves efficiently utilizing the emerging super resolution technique of upscaling the received low-resolution game frames from the server to high-resolution on mobile devices using DNN models in real-time, thus reducing the frame drops while improving the frame quality. (3) Finally, the third task involves a novel solution to enable a fast and compression-efficient encoding for the game streaming applications to further reduce the network bandwidth utilization as well as maximize the frame quality improvements. Overall, this dissertation lays the groundwork for enhancing game streaming on resource-constrained devices, promising faster, more energy-efficient, and bandwidth-optimized delivery of high-fidelity games, thus enabling a future of superior gaming experiences accessible to a wider audience.