Matt Stabeler

I managed to get a copy of the two InfoCom datasets and the Cambridge dataset  mentioned in the BubbleRAP paper, I could not find the Hong-Kong dataset. I parsed them into the correct format for LocationSim, and was able to genereate edge lists for all the connections, and have vizualised them below. Note: Node size represents betweeness centrality before edge removal. Edge width/colour represents connected time. Graphs below are for mobile nodes only; they do not include fixed nodes.

                 InfoCom05     InfoCom06     Cambridge
#iMotes           41              98             54
     Fixed        0               20             18
     Mobile       41              78             36
External          233             ?              ?
Duration         ~3 days        ~4 days         ~1 month (2 months in readme)

InfoCom 2006 edges removed where connected time less than 4 hours – (min = 1000ms max = 72 hours). InfoCom2006 gephi file.

InfoCom 2006 edges removed where connected time less than 2.5 hours – (min = 1000ms max = 46 hours). infocom2005 gephi file.

Cambridge with edges removed where connected time less than ~3 hours – (min = 1000ms max = ~39 hours). infocom2005 gephi file.

Cambridge Dataset with edges removed where connected time is less than ~3 hours - (min 1000ms max ~39 hours)

These graphs suggest that some community structure exists in the traces, and despite the duration of the InfoCom experiments being quite short, they may have more contact events due to the context of the experiment. (I wonder if the same patterns appear for broad scale deployments that appear in small area deployments?)

TODO:
Plot activity for duration of expermiments.
Parse SocialSensing Dataset…..

Activity every 15 mins for InfoCom2005 datset

Activity every 15 mins for InfoCom2006 datset

Activity every 12 hours for Cambridge dataset