This document discusses the influence of online games, traffic multiplexing, and router buffer on subjective quality. It presents the results of tests on multiplexing gaming traffic using different period sizes and router buffer configurations. The key findings are:
- Multiplexing can reduce bandwidth and packets per second by up to 30% and 35% respectively, at the cost of increased delay and jitter.
- Larger period sizes and router buffers increase delay more than smaller configurations. Multiplexing is limited by the bandwidth of around 1400kbps.
- Jitter also increases with multiplexing but is limited by the bandwidth, while smaller router buffers introduce less jitter and delay.
Influence of Online Games Traffic Multiplexing and Router Buffer on Subjective Quality
1.
2. INFLUENCE OF ONLINE GAMES
TRAFFIC MULTIPLEXING AND
ROUTER BUFFER ON
SUBJECTIVE QUALITY
GTC
Communication
Technologies Group
Jose Saldana
Julián Fernández-Navajas
José Ruiz-Mas
Eduardo Viruete Navarro
Luis Casadesus
University of Zaragoza, Spain
5. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
- Online games are getting very
popular in the last years
- First Person Shooters are the
ones with the tightest real-time
requirements
6. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
FPS use a client-server architecture
Uplink bandwidth limit
7. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
FPS use a client-server architecture
Server processing
capacity limit
8. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
FPS use a client-server architecture
Downlink bandwidth limit
9. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
“Shooting around the corner”
Jack
Wang
Wang
10. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
“Shooting around the corner”
Wang: Dead
Jack
Wang
Wang
11. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
“Shooting around the corner”
Wang: Dead
Jack
¡¡¡ #%$& !!!
Wang
12. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
Main Key Performance Indicators (KPI):
- Delay
- Jitter
- Packet loss
buffer
Internet
.
.
.
Users
Router
Game &
background
traffic
Game Server
13. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
Access networks: low-end routers, with
limitations in:
- Bandwidth
- Packets per second
Different implementations
buffer
Internet
.
.
.
Users
Router
Game &
background
traffic
Game Server
14. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
Scenarios where many users share a
connection: we can multiplex packets
from different players, and compress
headers
15. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Introduction
By multiplexing, we can mitigate two
problems:
- Bandwidth
- Packets per second
… at the cost of adding:
- Delay
- Jitter
17. II. Related Works
- Multiplexing real-time flows
- Subjective quality evaluation
- Influence of the router buffer
18. II. Related Works
- Multiplexing real-time flows
- Subjective quality evaluation
- Influence of the router buffer
19. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Multiplexing real-time
A multiplexer is introduced, and it also
compresses headers
delaymux
delayrouter
delaynetwork
IP network
router
.
.
.
MUX
DEMUX
Game Server
Players
IP
TCM
IP
20. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Multiplexing real-time
First done for VoIP (RFC4170, “TCRTP”):
payload
payload
ECRTP
...
ECRTP
RTP
UDP
IP
PPP Mux
PPP
L2TP
IP
Adapted for non-RTP flows:
Payload
Payload
Reduced Header
...
UDP
Reduced Header
IP
PPP Mux
PPP
L2TP
IP
21. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Multiplexing real-time
A period is defined, and all the packets
arrived are compressed and
multiplexed
Native
traffic
...
Multiplexed
traffic . . .
PE
PE
PE
PE
...
...
22. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Multiplexing real-time
Efficiency improvement
One IPv4/TCP packet 1500 bytes
η=1460/1500=97%
One IPv4/UDP/RTP packet of VoIP with two samples of 10 bytes
η=20/60=33%
One IPv4/UDP server-to-client packet of Counter Strike with 9 players
η=160/188=85%
Four IPv4/UDP client-to-server packets of Counter Strike
η=61/89=68%
One IPv4/TCM packet multiplexing four client-to-server Counter Strike packets
η=244/293=83%
saving
24. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Multiplexing real-time
Significant savings (Counter Strike)
Packets per second
600
20 players
15 players
500
10 players
5 players
pps
400
300
200
100
0
native
5
10
15
20
25
period (ms)
30
35
40
45
50
25. II. Related Works
- Multiplexing real-time flows
- Subjective quality evaluation
- Influence of the router buffer
26. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Multiplexing real-time
- E-Model: VoIP delay and packet loss
- FPS games: different studies
consider delay limits, and also
packet loss limits
- G-Model: MOS formula for Quake
IV, adapted from E-Model: delay
and jitter
27. II. Related Works
- Multiplexing real-time flows
- Subjective quality evaluation
- Influence of the router buffer
28. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Multiplexing real-time
- Buffer size
-
-
Bandwidth-delay product
Stanford model
Tiny buffer
Buffer implementation
-
-
Byte sized
Packet sized
Buffer limitations
-
Maximum pps amount
30. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Test methodology
- Simulation scenario:
-
-
Traces of gaming traffic
Background traffic
RTT delay: sum of the delays
delaymux
delayrouter
delaynetwork
IP network
router
.
.
.
MUX
DEMUX
Game Server
Players
IP
TCM
IP
31. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Test methodology
Jitter: Multiplexing increases it
stdevmux PE / 12
Different added delays
Native
traffic
...
Multiplexed
traffic . . .
PE
PE
PE
PE
...
...
32. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Test methodology
BW↓
Buffer
jitter↓
PE ↑
Mux
jitter↑
delaymux
delayrouter
delaynetwork
IP network
router
.
.
.
MUX
DEMUX
Game Server
Players
IP
TCM
IP
33. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Test methodology
Jitter has to be added in a proper way
(Network jitter is considered independent)
stdevmuxrouter stdevmux 2 stdevrouter2 cov mr
stdevtotal stdevmuxrouter2 stdevnetwork2
delaymux
delayrouter
delaynetwork
IP network
router
.
.
.
MUX
DEMUX
Game Server
Players
IP
TCM
IP
56. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Conclusions
- The mutual influence of the
buffer router and multiplexing
has been studied
- Small buffers are more adequate
to reduce delay and jitter
- Packet-sized buffers increase
packet loss, and multiplexing
alleviates this
57. CCNC 2012. DENVECT Workshop. Las Vegas Jan 14, 2012
Conclusions
- Multiplexing always saves
bandwidth
- In packet-sized buffers,
multiplexing does not improve
experienced quality
- Multiplexing is very interesting
when a limit in pps is present