- 1. May 31, 2007 Network Game Design: User Identification based
on Game-Play ActivityPatternsChun-Yang Chen, Academia SinicaLi-Wen
Hong, Academia Sinica ACM NetGames 2007
2. Motivation Password-based User Identity VulnerabilityAccount
hijacking (Identity Theft)Severity & prevalenceNo general
solution until the victim appearsAccount sharingIncrease the
difficulty of demographical studies of game Authors / Paper Title 2
3. User Identity: Current Solutions Digital signatureSmart
cardBiometrical signaturefingerprintvoicekeystroke Authors / Paper
Title3 4. Our Solution A novel biometric:Game-Play Activity
PatternsAuthors / Paper Title4 5. OutlineMotivationData
CollectionPlayer Activity AnalysisProposed SchemePerformance
EvaluationContribution & Future WorkAuthors / Paper Title 5 6.
Observation More RegularMore Unpredictable Motivation Data
Collection Player Activity Analysis Proposed Schemes Authors /
Paper Title6 7. Data Collection A MMORPG -- Angels LoveA commercial
game in Taiwan40 thousands of players onlineThe player activity
logs we useTrace period of 3 days287 randomly chosen accountsRemove
logs shorter than 200 minutes Motivation Data Collection Player
Activity Analysis Proposed Schemes Authors / Paper Title7 8.
Definitions Active periodAn active period of a game character is
defined as a timeinterval (t1 , t 2 ) which the character
continuously moves, inwith a tolerance of discontinuity up to 1
second.Idle periodAn idle period of a game character is defined as
a timeinterval (t1 , t 2 ) which the character has no movements,
inwhere t 2 t1 second. 1 Data Collection Player Activity Analysis
Proposed Schemes Performance Evaluation Authors / Paper Title8 9.
Data Summary Player # :DataActivity Active Idle 287
LengthRatePeriodPeriod0.355%7 hr 3 sec 7 seccycle/min 2.28 50%51
hr6 sec 18 seccycle/min 5.12 95%67 hr9 sec181 seccycle/min Data
Collection Player Activity Analysis Proposed Schemes Performance
Evaluation Authors / Paper Title9 10. Distribution of Active / Idle
Period 4Data Collection Player Activity Analysis Proposed Schemes
Performance Evaluation Authors / Paper Title10 11. Idle time is
much more diversethan active time Data Collection Player Activity
Analysis Proposed Schemes Performance Evaluation Authors / Paper
Title11 12. Average Active Time v.s. Idle Time Players active/idle
patterns can be very different candidate features for user
identification Data Collection Player Activity Analysis Proposed
Schemes Performance Evaluation Authors / Paper Title12 13. Why
choosing idle time ratherthan active time Idle time distribution
captures more variability Idle time process has smaller degree of
auto- correlations Data Collection Player Activity Analysis
Proposed Schemes Performance Evaluation Authors / Paper Title13 14.
Idle Time Distribution ofRandom Players Data Collection Player
Activity Analysis Proposed Schemes Performance Evaluation Authors /
Paper Title14 15. KL Distances ITD: Idle time distribution RET:
Relative Entropy Testrelative entropy between two ITDsbased on the
KL distance KL distance: Kullback-Leibler distance P(i ) DKL ( P ||
Q) = P(i ) log i Q(i )DSKL ( P || Q) = DSKL (Q || P) = DKL ( P ||
Q) + DKL (Q || P) Player Activity Analysis Proposed Schemes
Performance Evaluation Authors / Paper Title15 16. KL distances of
Players Player Activity Analysis Proposed Schemes Performance
Evaluation Authors / Paper Title16 17. Identification
Scheme:Consistency Test Perform consistency testKLD: distribution
of KL distance2 KLDs for each playerKLDs are tested by two-sided
Wilcoxon test0.95 Authors / Paper Title 17 18. Identification
Scheme:Discriminability Test Perform discriminability testKLDi,j :
distribution of KL distances between ni ITDs of player i& nj
ITDs of player j KLDs are tested by one-sided Wilcoxon test Authors
/ Paper Title18 19. Factor Consideration Consideration: effect of
the detection time & the history sizeTrec: how long the player
history kept in database (in minutes)Tobs : the detection time once
the player log in (inminutes)Proposed Schemes Performance
Evaluation Contribution & Future Work Authors / Paper Title19
20. Evaluation Result Proposed Schemes Performance Evaluation
Contribution & Future Work Authors / Paper Title20 21.
Performance Evaluation Effect of Trecmean of activity cycles is one
minuteTrec = idle timesassuming one million user accounts, Trec =
200 minutes, eachidle time uses 4 bytes storage space = 800
MBEffect of Tobsassuming 10,000 players are online, Tobs = 20
minutesmain memory = 0.8 MBPlayer Activity Analysis Proposed
Schemes Performance Evaluation Authors / Paper Title 21 22.
Contribution & Future Work Contribution: Propose the RET scheme
for user identification from the aspect of idle time.With a
20-minute detection time period given a 200-minutehistory size
achieve higher than 90% accuracy.Future PlanCut down the detection
timeUtilizing more aspects of game-play activities.Analyzing from
the way users control the character. Proposed Schemes Performance
Evaluation Contribution & Future Work Authors / Paper Title22
23. Thank you! Chun-Yang Chen Li-Wen Hong ACM NetGames 2007