1
M MNe t Chih-Ming Chen 1 , Kuan-Ta Chen 2 , and Polly Huang 1 1 National Taiwan University 2 Academia Sinica Lab Motivation Motivation There is no consensus on protocols for MMORPGs MMORPGs requirements Low transmission latency No unexpected “lags” Multistreaming Put different message types into separate streams e.g., move message, attack message, talk message Optional ordering Certain message types do not require inorder transport e.g., move message, attack message Optional reliability Certain message types do not require reliability e.g., move message Quantifying the Effect of Content Quantifying the Effect of Content based Transport based Transport Strategies for Online Role Playing Games Strategies for Online Role Playing Games Quantify the effect of contentbased transport strategies Evaluate existing transport protocols TCP, UDP, DCCP, SCTP Propose and evaluate three contentbased transport strategies usingnetwork simulations Objective Objective http://mmnet.iis.sinica.edu.tw Multimedia Networking and Systems Lab Institute of Information Science, Academia Sinica Protocol MMORPGs TCP World of Warcraft, Lineage I/II, Guild Wars, Ragnarok Online, Anarchy Online, Angel’s Love UDP EverQuest, City of Heroes, Star Wars Galaxies, Ultima Online, Final Fantasy XI TCP/UDP Dark Age of Camelot Users’ action trace of Angel’s Love Real Real - - Life Game Traces Life Game Traces User ID Actions (M: Move, A: Attack, T: Talk) 10159 MMMMMMMMMMMMMAMMAMMMMMMAMMMMMMMMMM 12454 MMAAMAMMMMMAAMAMAMAMAMAMAMAMAMAMAMAMA 16728 MMMMMMMMMMMMMMMMMMMMMMAMMTTTTTTTTTTTT Message Type In-Order Delivery Reliability Move Attack Talk Strategy Multi-Streaming Optional Ordering Optional Reliability MRO MR M Performance Evaluation Performance Evaluation Average endtoend transmission latencies Average endtoend delay jitters (standard deviation of delays) Link Bandwidth Propagation Delay Server <--> Router 600 Kbps 70 ms Router <--> Router 600 Kbps 70 ms Router <--> Client 64 ~ 128 Kbps 70 ms Router <--> Traffic Node 64 ~ 128 Kbps 70 ms Summary Summary Simulation Setup Simulation Setup Transport Strategies Transport Strategies Tracedriven network simulation using ns2 simulator Fishbone topology 1 game server 2 network routers n game clients m cross traffic pairs 20 40 60 80 100 120 140 180 200 220 240 260 280 300 Client Traffic Client # Mean delay (ms) TCP UDP SCTP P_MRO P_MR P_M 20 40 60 80 100 120 140 200 300 400 500 600 Server Traffic Client # Mean delay (ms) TCP UDP SCTP P_MRO P_MR P_M 20 40 60 80 100 120 140 0 50 100 150 200 250 Client Traffic Client # Mean jitter (ms) TCP UDP SCTP P_MRO P_MR P_M 20 40 60 80 100 120 140 0 100 200 300 400 500 600 Server Traffic Client # Mean jitter (ms) TCP UDP SCTP P_MRO P_MR P_M Bandwidth & propagation delay 11 pairs cross traffic with 500 Kbps sending rate P_MRO P_MR P_M Client Traffic Protocol Score 0 20 40 60 80 100 0 2 0 0 55 75 70 Clients 140 Clients P_MRO P_MR P_M Server Traffic Protocol Score 0 20 40 60 80 100 2 8 10 33 45 69 70 Clients 140 Clients P_MRO P_MR P_M Client Traffic Protocol Score 0 20 40 60 80 100 0 1 1 0 69 79 70 Clients 140 Clients P_MRO P_MR P_M Server Traffic Protocol Score 0 20 40 60 80 100 1 5 5 28 46 73 70 Clients 140 Clients Improvement of Delay Improvement of Delay Jitter Server CN 1 CN 2 TN 1 CN i TN 2 TN j CN i+1 CN i+2 TN j+1 CN n TN j+2 TN m ... ... ... ... Client Node (CN) Traffic Node (TN) Client to Server Server to Client Protocol Delay Jitter Delay Jitter TCP NA NA NA NA SCTP DCCP (TCP-like) ★★★★ ★★★★ ★★★ ★★★ DCCP (TFRC) UDP ★★★★ ★★★★ ★★★★ ★★★★ P MRO P MR ★★ ★★ P M ★★★ ★★★ ★★★ ★★ Description: NA denotes incomparable, denotes similar, denotes worse, denotes better, ★★ denotes much better, ★★★ denotes good, ★★★★ denotes very good, and ★★★★ denotes excellent.

Quantifying the Effect of Content-based Transport Strategies for Online Role Playing Games

Embed Size (px)

DESCRIPTION

The QoS requirements of game messages may differ because of the latter’s intrinsic characteristics. In this paper, we propose three content-based strategies for quantifying the effects of different QoS levels. The strategies assign appropriate QoS requirements for game messages based on our analysis of Angel’s Love action logs. We evaluate several transport protocols, including TCP, UDP, SCTP, DCCP, and our content-based transport protocol using the action logs of Angel’s Love. Through simulations, we quantify the performance of our content-based strategies. The results show that the strategies incur much lower end-to-end delay and end-to-end jitter than existing transport protocols.

Citation preview

Page 1: Quantifying the Effect of Content-based Transport Strategies for Online Role Playing Games

M

MNetChih-Ming Chen1, Kuan-Ta Chen2, and Polly Huang1

1National Taiwan University 2Academia SinicaLab

MotivationMotivationMotivation

There is no consensus on protocols for MMORPGs

MMORPGs requirementsLow transmission latencyNo unexpected “lags”

Multi‐streaming• Put different message types into separate streams• e.g., move message, attack message, talk messageOptional ordering• Certain message types do not require in‐order transport• e.g., move message, attack messageOptional reliability• Certain message types do not require reliability• e.g., move message

Quantifying the Effect of ContentQuantifying the Effect of Content‐‐based Transportbased TransportStrategies for Online Role Playing GamesStrategies for Online Role Playing Games

Quantify the effect of content‐based transport strategies

Evaluate existing transport protocols

• TCP, UDP, DCCP, SCTPPropose and evaluate three content‐based transport strategies  usingnetwork simulations

ObjectiveObjectiveObjective

http://mmnet.iis.sinica.edu.tw

Multimedia Networking and Systems LabInstitute of Information Science, Academia Sinica

Protocol MMORPGsTCP World of Warcraft, Lineage I/II, Guild Wars, Ragnarok

Online, Anarchy Online, Angel’s LoveUDP EverQuest, City of Heroes, Star Wars Galaxies, Ultima

Online, Final Fantasy XITCP/UDP Dark Age of Camelot

Users’ action trace of Angel’s Love

Real-Life Game TracesRealReal--Life Game TracesLife Game Traces

User ID Actions (M: Move, A: Attack, T: Talk)10159 MMMMMMMMMMMMMAMMAMMMMMMAMMMMMMMMMM12454 MMAAMAMMMMMAAMAMAMAMAMAMAMAMAMAMAMAMA16728 MMMMMMMMMMMMMMMMMMMMMMAMMTTTTTTTTTTTT

Message Type In-Order Delivery ReliabilityMoveAttack √

Talk √ √

Strategy Multi-Streaming Optional Ordering

Optional Reliability

MRO √

MR √ √

M √ √ √

Performance EvaluationPerformance EvaluationPerformance EvaluationAverage end‐to‐end transmission latencies

Average end‐to‐end delay jitters (standard deviation of delays)

Link Bandwidth Propagation DelayServer <--> Router 600 Kbps 70 msRouter <--> Router 600 Kbps 70 msRouter <--> Client 64 ~ 128 Kbps 70 msRouter <--> Traffic Node 64 ~ 128 Kbps 70 ms

SummarySummarySummarySimulation SetupSimulation SetupSimulation Setup

Transport StrategiesTransport StrategiesTransport Strategies

Trace‐driven network simulation using ns‐2 simulatorFishbone topology• 1 game server • 2 network routers• n game clients • m cross traffic pairs

20 40 60 80 100 120 140

180

200

220

240

260

280

300

Client Traffic

Client #

Mea

n de

lay

(ms)

TCPUDPSCTP

P_MROP_MRP_M

20 40 60 80 100 120 140

200

300

400

500

600

Server Traffic

Client #

Mea

n de

lay

(ms)

TCPUDPSCTP

P_MROP_MRP_M

20 40 60 80 100 120 140

050

100

150

200

250

Client Traffic

Client #

Mea

n jit

ter

(ms)

TCPUDPSCTP

P_MROP_MRP_M

20 40 60 80 100 120 140

010

020

030

040

050

060

0

Server Traffic

Client #

Mea

n jit

ter

(ms)

TCPUDPSCTP

P_MROP_MRP_M

Bandwidth & propagation delay

11 pairs cross traffic with 500 Kbps sending rate

P_MRO P_MR P_M

Client Traffic

Protocol

Sco

re

020

4060

8010

0

0 2 0 0

55

75

70 Clients140 Clients

P_MRO P_MR P_M

Server Traffic

Protocol

Sco

re

020

4060

8010

0

28 10

33

45

69

70 Clients140 Clients

P_MRO P_MR P_M

Client Traffic

Protocol

Sco

re

020

4060

8010

0

0 1 1 0

69

79

70 Clients140 Clients

P_MRO P_MR P_M

Server Traffic

Protocol

Sco

re

020

4060

8010

0

15 5

28

46

73

70 Clients140 Clients

Improvement of Delay Improvement of Delay Jitter

Server

CN1

CN2

TN1

CNi

TN2TNj

CNi+1

CNi+2

TNj+1

CNn

TNj+2TNm

. . .  . . . 

. . .  . . . 

Client Node (CN)Traffic Node (TN)

Client to Server Server to ClientProtocol Delay Jitter Delay JitterTCP NA NA NA NASCTP ≈ ≈ ★ ★

DCCP (TCP-like) ★★★★ ★★★★ ★★★ ★★★

DCCP (TFRC) ☆ ☆ ☆ ☆

UDP ★★★★ ★★★★ ★★★★ ★★★★

PMRO ≈ ≈ ★ ★

PMR ≈ ≈ ★★ ★★

PM ★★★ ★★★ ★★★ ★★

Description: NA denotes incomparable, ≈ denotes similar, ☆ denotes worse, ★ denotes better, ★★ denotes much better, ★★★ denotes good, ★★★★ denotes very good, and ★★★★ denotes excellent.