Introduction Mining Second Life Measurement Methodology Results Conclusion Mining Second Life: Characterizing User Mobility in a Popular Virtual World Chi-Anh La - Pietro Michiardi ACM WOSN 2008 Seattle, WA, U.S.
Introduction Mining Second Life Measurement Methodology Results Conclusion Outline of the talk Introduction 1 Mining Second Life 2 Measurement Methodology 3 Results 4 Conclusion 5
Introduction Mining Second Life Measurement Methodology Results Conclusion Characterizing human mobility: Objectives of this work: Define a novel methodology to carry out experiments on human mobility with the following goals: Affordable experiments No logistic organization Wireless technology independent Scalability of experiments
Introduction Mining Second Life Measurement Methodology Results Conclusion Related works Objectives of prior works: Build mobility models from traces Performance evaluation of forwarding strategies in DTNs Chaintreau et. al. : IEEE Trans. Mobile Computing 2007 Karagiannis et. al. : ACM Mobicomm 2007 Rhee et. al. : IEEE Infocom 2008
Introduction Mining Second Life Measurement Methodology Results Conclusion Related works: Experimental Methodology Select hardware → exhausting task Neighbor discovery → hard for wifi in ad-hoc mode Prepare / finalize the experiment → logistic problems
Introduction Mining Second Life Measurement Methodology Results Conclusion Related works: Restrictions Available traces are difficult to use (and debug) Experiments are bound to specific wireless hardware In general, only “temporal” information is available GPS-based experiments only for out-door scenarios Number of participants to experiments is fixed
Introduction Mining Second Life Measurement Methodology Results Conclusion The idea Exploit Virtual Worlds Networked Virtual Environment are a tremendously popular con- cept of on-line communities: User interaction is synchronous Contrast with Social-Networking applications such as FaceBook: asynchronous interaction In this work we use Second Life and capture user interaction as well as user spatial distribution.
Introduction Mining Second Life Measurement Methodology Results Conclusion Our Playground: Second Life Second Life architecture: Flat, Earth-like world simulated on a large server farm World is divided into 256x256 m “lands”, one server per land → Limitation on number of concurrent users on each land Each land has attributes: private public sandbox → Limitations on user-generated content deployment
Introduction Mining Second Life Measurement Methodology Results Conclusion Monitoring Architectures Measurements in Second Life can be approached under different angles Use Second Life to build and deploy monitoring probes Use Second Life to mimic real world experiments System approach: connect to Second Life and get data We built a lightweight client wich crawls a selected land Input : Valid Login/passwd Target Land Measurement granularity Measurement duration Output: Anonymized user ID ( x , y , z ) of every user on the target land every τ seconds
Introduction Mining Second Life Measurement Methodology Results Conclusion The Crawler Approach Observations: The crawler is a user → should not introduce bias in experiments One crawler per land is sufficient All users concurrently connected to the target land can be tracked: we override a method used to build maps Multiple lands can be tracked using an “army” of crawlers Limitation: maximum number of concurrent users
Introduction Mining Second Life Measurement Methodology Results Conclusion Measurement Methodology We present results for the following lands: Open Spaces: Apfel Land: a german-speaking arena for newbies Island of View: Valentine’s day event Confined areas: Dance Island: a virtual discotheque Note: Selecting lands is a tedious manual exercise Automate the process
Introduction Mining Second Life Measurement Methodology Results Conclusion Using SecondLife Traces How do we use the traces? Using the coordinates of users connected to a target land we create several snapshots of radio networks Given a communication range r , a link between two users u i , u j exists if their distance d ( u i , u j ) ≤ r We build snapshots every measurement interval τ = 10 sec r ∈ { r b , r w } , where r b = 10 m (bluetooth) and r w = 80 m (WiFi at 54 Mbps)
Introduction Mining Second Life Measurement Methodology Results Conclusion Metrics Temporal: Contact Time Inter Contact Time Spatial: Node degree distribution Network diameter Clustering Coefficient Zone occupation Mobility: Cumlative traveled distance Login time
Introduction Mining Second Life Measurement Methodology Results Conclusion Results: Some Numbers 24-hours traces Apfel Land: Unique visitors: 1568 Average concurrent users: 13 Dance Island: Unique visitors: 3347 Average concurrent users: 34 Isle of View: Unique visitors: 2656 Average concurrent users: 65
Introduction Mining Second Life Measurement Methodology Results Conclusion Results: Temporal Analysis (1) Contact Time CCDF, r=80m Inter − Contact Time CCDF, r=80m 1 1 0.5 0.5 0.1 0.1 1 − F(x) 1 − F(x) Apfelland Apfelland Dance Dance Isle Of View Isle Of View 1 2 3 4 5 1 2 3 4 5 10 10 10 10 10 10 10 10 10 10 Time (s) Time (s) Contact Time = transfer Inter Contact Time = time opportunities between to wait before a pair meets users again Large values are good Large values are supposedly bad
Introduction Mining Second Life Measurement Methodology Results Conclusion Results: Trip Characteristics Travel Length CDF Travel Time CDF 1 1 0.9 0.9 0.8 0.8 0.6 0.6 F(x) F(x) 0.5 0.5 0.4 0.4 0.2 Apfelland 0.2 Apfelland Dance Dance 0.1 0.1 Isle Of View Isle Of View 0 0 0 5000 10000 15000 20000 0 500 1000 1500 2000 2500 Time (s) Length (m) Users do not exercise a lot! Max on-line time ∼ 4 h Closed vs. open spaces 90-th perc. on-line < 1 h Our explanation: Quite obvious (and similar to real world): users do not move when they chat!
Introduction Mining Second Life Measurement Methodology Results Conclusion Results: Spatial Distribution Zone Occupation CDF, L=20m Zone Distribution, Isle Of View, L=20m 1 Number of users per cell 8 6 0.95 4 2 F(x) 0.9 0 200 200 100 0.85 100 Apfelland 0 0 Y X Dance Isle Of View 0.8 0 5 10 15 20 25 Number of users per cell Zone Distribution, Dance, L=20m Number of users per cell 8 Not a uniform distribution 6 Most of the users are 4 2 grouped 0 200 200 Closed vs. open spaces 100 100 0 Y 0 X
Introduction Mining Second Life Measurement Methodology Results Conclusion Concluding remarks Novel approach to study mobility Do real people walk like avatars? Beyond mobility analysis Epidemiology Sociology Virtual playground to test applications
Introduction Mining Second Life Measurement Methodology Results Conclusion Thank you! Need traces? Contact: Pietro.Michiardi@eurecom.fr Web: www.eurecom.fr/ ∼ michiard
Recommend
More recommend
