CPUH-Research Journal: 2015, 1(2), 19-22

ISSN (Online): 2455-6076

http://www.cpuh.in/academics/academic_journals.php

Cloud Gaming: Architecture and Quality of Service

Anmol Gupta1* and Kamlesh Dutta1
1
Department of Computer Science & Engineering, NIT, Hamirpur (H.P.), INDIA
*
Correspondance Email: bratanmol@gmail.com, kdnith@gmail.com

ABSTRACT: Cloud Gaming can be defined as the offloading of game logic to the cloud server and rendering the
game scenes as a video stream transfer to the client. The thin client sends the user commands to the remote server
which executes the game logic and sends back the changed game scene to the client in the form of a video, which is
decoded at the client side. The scheme helps people to play games on their computer system irrespective of the
specification of their systems as the requirements for playing the game are fulfilled with the help of a good enough
server residing on the cloud. Cloud gaming also helps to overcome the traditional gaming issues such as incompati-
bility and portability. Cloud gaming differs from online gaming in executing the main game logic, which is done at
the server side in former and at the client side in latter. In this paper, the different aspects of the Cloud Gaming
System (CGS) are discussed to illustrate the various advantages and power that this concept possesses. The cloud
gaming concept, architecture of a cloud system, and Quality of Service (QOS) parameters are discussed in detail. A
cloud gaming system should aim at providing the best response time and low bandwidth consumption. The future
of cloud gaming seems to be bright as some cloud gaming systems like Gaikai and On Live are gaining popularity
at a commendable pace.

Keywords: Cloud Gaming; cloud server and Architecture.

INTRODUCTION: Cloud Gaming is defined as the on it. While in cloud gaming architecture the execu-
use of remote resources in order to play games on the tion of program logic is done at server side2.
local system. Generally, the UI input events are transmit-
ted with the help of a thin client, and then the game logic Why Cloud Gaming is a good idea? Consider the
executes on the remote server which in turn, is reflected case of a First Person Shooter (FPS) game “Battlefield
back to the client in the form of a video stream which 3”. The minimum requirements for playing this game
gets played with the help of thin client. is – Quad Core Processor, 4 GB RAM, 20 GB storage
space, and a graphics card with at least 1GB RAM
Computer games are graphics intensive and hence (e.g., NVIDIA GEFORCE GTX 560 or ATI RADEON
there is a demand for new hardware which can meet 6950), which alone costs more than $500. Needless to
these high-end demands1. The limitations affect the say that the recent mobile systems, tablets or even
individual game player who has to change his hard- some of the PCs don't meet this requirement. Also
ware anytime a new incompatible version of a game is mobile terminals don't have the same architecture of a
released. Also, there is hardware/software incompati- PC, have power limitations, distinct operating system,
bility and the additional time required in setting up the and many more limitations. All these are overcome by
games. using Cloud Gaming. Cloud Gaming also offers an
In cloud gaming, the user can access the hardware of added advantage of better Digital Rights Management
the cloud systems thus eliminating the need of recon- (DRM) as the hardware being used is controlled by
figuring his/her own hardware, installing the game or the owner of the cloud system rather than the user [3].
worrying about software/hardware incompatibility. Cloud Gaming being such a computationally intensive
Another significant advantage is that the game can be process requires many technologies to work in unison
played on any system such as PC, mobile device and so that the user can't fathom the remote execution of
even on TVs with Set-top boxes provided you have a the main game logic. In this paper, a detailed analysis
good enough internet connection. of the various technologies used in providing a pleas-
ant cloud gaming experience to the end user, is pre-
It is noteworthy that Cloud Computing is different sented.
from online gaming. “Traditional” online games also
take advantage of remote resources on the cloud but The paper is divided into following sections. After the
the execution of game logic is done at client side only, introduction in section I, section II provides the gen-
while the servers are there for handling the consisten- eral architecture of a cloud gaming system. Section III
cy between the states of various games being played comprises of the various Quality of Service (QOS)
parameters used for measuring the End User Experi-

Proceedings of the National Conference on Recent Innovations in Science and Engineering (RISE-2016) 19
 [(CPUH-Research Journal: 2015, 1(2), 19-22) Cloud Gaming: Architecture and Quality of Service ]

ence. Finally section IV concludes the paper with a service to the thin client which then decodes the
look ahead at the future of cloud gaming. stream and play the video frames to the client [3].
General Architecture of a Cloud Gaming System There are several design alternatives which can be
(CGS): Generally there can be three different types of considered to provide different implementations of a
remote rendering Real time Systems. (i) 3D Graphics CGS.
Streaming4 & 7, (ii) video streaming5 & 6, and (iii) video
1) The way the existing game software is modified
streaming with post-rendering operations14 .
and run on the server
In 3D Graphics system the cloud server sends the 2) The way the game screen is encoded (on the serv-
graphics related commands to the client which then er) and decoded (on the client)
interprets it and renders the scene accordingly while in 3) The way the encoded game screen is streamed to
Video Streaming the server is responsible for render- the client.
ing the 3D commands, converting them to 2D and 4) The way short-term network instability is handled
then sending the video stream to the client. The third to maintain the game’s responsiveness and graphic
system comes in between the first two where the quality.
heavy work of rendering the 3D graphics is done on
Various QOS Parameters: The effectiveness of a
server while some low processor intensive work is
CGS can be measured from many point of views,
completed on the client side via Thin Client.
from the service provider's view, the allocation of
There have been many thin client architectures pro- resource is important while from the end user's per-
posed for distributed gaming systems [4-7]. They can be spective the metrics which affect gaming experience
divided into two categories – instruction based sys- are important. We can also quantify the QOS from the
tems and image based systems. The main difference time scale, it can be larger time-scale across multiple
between the two is that, in instruction based systems, gaming sessions and smaller across individual game
only the instructions for generating the graphics corre- session. Here we will focus only upon small time-
sponding to a control event are sent over the network scale systems because most CGSs uses only single
while in image-based systems all the computationally Virtual Machine without any offloading to serve each
intensive rendering of the game scene is done on the client.
server side and is sent over the network in the form of
The main metrics that are focused upon:
a video stream. All the CGSs use image based thin
1. Traffic Characteristics: The amount of bandwidth
client architectures only because they don't require
utilized in a single game session. It also includes
client to have the computing resources which is the
payload size and packet rate (both uplink and down-
whole selling point of Cloud Gaming.
link).
Here we discuss the general image based architecture 2. Latency: It's probably the most important criterion
of a CGS. for measuring the performance of CGSs. It can be
defined as the response time of a system and it com-
Thin Client: It consists of a User Interaction module
prises all the individual latencies incurred by differ-
and a video decoder. The User Interaction module is
ent components.
responsible for capturing all the control movements
3. Graphics Quality: The quality of the images/videos
performed by the end user with the help of mouse,
streamed over the network is very important for the
keyboard or any other input device.
Quality of Experience (QOE) for the user. Measure-
The video decoder plays the video being streamed by ments also include the changes in quality over vary-
the server in response to the moves made by the play- ing network conditions. Streaming quality is general-
er. ly measured with the metric Frame rate (FPS).
Cloud Gaming Platform: It consists of four modules Graphic Quality can be measured by Peak Signal to
Noise Ratio (PSNR) or Structural Similarity Index
namely Thin Client Interaction, Game Logic, GPU
Renderer and Video Encoder. Thin Client Interaction Method (SSIM) metric9.
module handles all the client commands and it con- Latency quantifies the responsiveness of game playing
verts the messages sent over the network into appro- and is a measure of Response Delay which is the
priate game actions. These are interpreted by the game amount of time between a user sends his/her command
logic module in game world and then the GPU will and the time the corresponding game frame is dis-
actually render the scene which then gets compressed played to the user.
by the encoder and then sent by the Video Streaming

Proceedings of the National Conference on Recent Innovations in Science and Engineering (RISE-2016) 20
 [(CPUH-Research Journal: 2015, 1(2), 19-22) Cloud Gaming: Architecture and Quality of Service ]

Response Delay (RD) comprises of four individual lay (PD), Game Delay (GD) and Playout Delay (OD)2.
delays namely Network Delay (ND), Processing De-

User Interaction Model Cloud Server

User sends Commands

User Game Logic
Interaction Executer

Game Scene
Changes

Video Sends back the new stream Video

Decoder Renderer

Figure 1: General Framework of a CGS

Playout Delay (OD): Time for client to receive, de-
Network Delay (ND): The round trip time of the net-
code and play the current frame.
work, it actually measures the time taken by a client's
command to reach the server and the time taken by the RD = ND + PD + GD + OD
game screen to come to the client.
ND can be measured with the help of ICMP pings or
Processing Delay (PD): Time taken by the server to Wireshark or any other such network tools while GD
receive and process the user's command. It also in- is game dependent. Measuring PD and OD is not
cludes time to encode and packetize the current frame straightforward because they occur internally at the
for the client. server and the client respectively.
Game Delay (GD): Time taken by the game software As mentioned that RD is an important factor in decid-
to process the user's command and generate the corre- ing the quality of the CGS, various studies have been
sponding game frame. This is usually considered to
performed on the delay tolerance for different kind of
remain unchanged in the cloud environment and games. The results can be summarized as [10]:
hence GD in standalone gaming equals GD in Cloud
Gaming.
Table 1: Delay Tolerance in Traditional Gaming.
Example Game Type Perspective Delay Thresh-
First Person Shooter First Person 100 ms
Role Playing Game Third- 500 ms
Real Time Strategy Omnipres- 1000 ms

CONCLUSION: This paper compiles the various as- thin client. The future of cloud computing seems bright
pects of cloud computing and discusses some of it’s as network latencies are becoming shorter with the ad-
aspects at length. The platform is gaining much populari- vent of technologies such as 3G/4G making their mark.
ty and is particularly powerful in bringing the graphics Also, the techniques used in cloud gaming can serve for
and processing heavy games to the mobile platforms, other industries too, such as video streaming and
which in comparison to a general computer system, has graphics rendering.
less resources. Some of the shortcomings of cloud gam- On Live and Gaikai are the two mainstream commer-
ing are high network latency and the effectiveness of cial platforms which are successful in the market

Proceedings of the National Conference on Recent Innovations in Science and Engineering (RISE-2016) 21
 [(CPUH-Research Journal: 2015, 1(2), 19-22) Cloud Gaming: Architecture and Quality of Service ]

while open source platforms are emerging slowly. One 12. Giesen, F., Schnabel, R. and Klein, R. 2008.
such open source platform is Gaming Anywhere. Augmented compression for server-side render-
ing. In Proc. of International Fall Workshop on
As there are many design alternatives available, it is
Vision, Modeling, and Visualization (VMV’08).
not yet decided which design system offers the best
quality of service and an empirical study is needed to
help the decision.

REFERENCES:
1. Claypool, M. 2009. Motion and scene complexity
for streaming video games,” in Proceedings of the
4th International Conference on Foundations of
Digital Games. ACM. 34–41.
2. Chen, K. T. 2013. On the quality of service of
cloud gaming systems. IEEE transactions on mul-
timedia.
3. Shea, R. Cloud Gaming: Architecture and Per-
formance.
4. Jurgelionis, A., Fechteler, P., Eisert, P., Bellotti,
F., David, H., Laulajainen, J. P., Carmichael, R.,
Poulopoulos, V., Laikari, A., Per ̈al ̈a, P., Gloria,
A. De and Bouras, C. 2009. Platform for distrib-
uted 3D gaming,” Int. J. Comput. Games
Technol., 1:1-15.
5. Winter, D. De, Simoens, P., Deboosere, L., Turck,
F. De, Moreau, J., Dhoedt, B. and Demeester, P.
2006. A hybrid thin-client protocol for multimedia
streaming and interactive gaming applications,” in
Proceedings of ACM NOSSDAV 2006. ACM,
15:1–6.
6. Holthe, I., Mogstad, O. and Rønningen, L. A.
2009. Geelix livegames: Remote playing of video
games,” in Proceedings of IEEE CCNC 2009.
IEEE Press. 758–759.
7. Eisert, P. and Fechteler, P. 2008. Low delay
streaming of computer graphics,” in Proceedings
of IEEE ICIP 2008. 2704–2707.
8. Nave, I., David, H., Shani, A., Tzruya, Y.,
Laikari, A., Eisert, P. and Fechteler, P. 2008.
Games@Large graphics streaming architecture,”
in IEEE International Symposium on Consumer
Electronics 2008.
9. Wang, Z., Bovik, A., Sheikh, H. and Simoncelli,
E. 2004. Image quality assessment: from error vis-
ibility to structural similarity. IEEE Transactions
on Image Processing 13: 600–612.
10. Claypool, M. and Claypool, K. 2006. Latency and
player actions in online games. Communications
of the ACM, 49: 40– 45.
11. Shi, S., Hsu, C. Nahrstedt, K. and Campbell, R.
2011. Using graphics rendering contexts to en-
hance the real-time video coding for mobile cloud
gaming. In Proc. of ACM Multimedia’11. 103–
112.

Proceedings of the National Conference on Recent Innovations in Science and Engineering (RISE-2016) 22