Home > Uncategorized > LBR Location Analysis – MIT Dataset (Oct)

LBR Location Analysis – MIT Dataset (Oct)

December 8th, 2010 Leave a comment Go to comments

I wanted to see what the real picture was when using cell towers for location, so for a run of LBR on the MIT Dataset for October, I collected details about every message hop. I compared the timestamps for the hops to the cell tower information in the database, to find out which cell tower each node involved in the message hop was recorded as using.

Colocated: (same cell) 2247

Not-Colocated: (different cells) 15688

One sided: (only one sees cell) 13452

No data: (no cells seen) 3211

Count: (total number of hops evaluated) 31387

These results are slightly concerning, as roughly half of the hops between nodes happened when nodes reported different locations (cell towers ID),  10% of hops did not record any cell tower information at all, and in 42% of hops, only one node reported a location. This seems to demonstrate that we cannot rely on cell tower information to give us an accurate idea of the co-location of people. However, it might still prove useful to use the cell tower location for generating general mobility statistics, which could be used as routing metrics. Davide mentioned he had some ideas about these sort of statistics.

Categories: Uncategorized
  1. No comments yet.
  1. No trackbacks yet.