Skip to main navigation Skip to search Skip to main content

Action-aware combat model for efficient video compression of massively multiplayer online role-playing games on cloud gaming platforms

  • Sardar Basiri

Student thesis: Master's thesisMaster in Engineering: Systems Engineering

Abstract

Cloud gaming allows users to play new, high-quality games remotely without needing a powerful device. Users can connect remotely with their thin-client devices (e.g., tablets and smartphones) to send their control input and, in response, to receive a video from the cloud gaming platform. The video encoding and the transmission of the compressed video over the Internet are the main factors affecting the end-to-end latency, starting from the user’s input to the reception of the resulting video. This latency should be kept minimal in order to improve the game’s responsiveness and the quality of experience (QoE) of the player. The other important factor affecting the player’s experience is the quality of the video stream. Since better video quality requires more bandwidth (controlled by the bit rate) from the player, the challenge addressed in this thesis is to maximize the quality of the video stream with the same video bit rate or, equivalently, reducing the video bit rate for the same video quality. Our main focus is on massively multiplayer online role-playing games (MMORPGs) running on cloud gaming platforms. MMORPGs are the genre of online games where a huge number of players can interact in a vast virtual environment. The most crowded area is the combat area, where players use different weapons and skills associated with their game characters to interact with each other. We present an ARCODE for MMORPGs on cloud gaming platforms. The model determines the importance of each object in each state of the game, considering the properties of various actions that could be performed by any object in the combat area. Based on the importance of each object in each state of the game, the model decides how frequently the position of the object in the scene should be updated. By reducing the smoothness of movement of objects based on their significance to the player in each state of the game, fewer bits are required by the encoder to code the video frames with the same visual quality as the traditional approach. The model can be used with most MMORPGs in order to improve the quality of the compressed video when using the same amount or fewer bits. The advantage of our approach is that the coder-decoder (CODEC) used for the cloud gaming stream is considered as a black box. No change is applied to the encoder itself or to the scripts and the inputs of the encoder, which improves the portability and compatibility of our solution with established cloud gaming platforms. A wide selection of test scenarios was considered for the evaluation and to show the excellent performance of our model. Our experimental results indicate that for the same visual quality as the traditional approach, ARCODE reduces the video bit rate between 10% and 41%.
Date7 Jul 2021
Original languageAmerican English
Awarding Institution
  • École de technologie supérieure
SupervisorKaiwen Zhang (Supervisor) & Stéphane Coulombe (Co-supervisor)

Cite this

'