Matt Stabeler

Talked again about location clustering and proximity methods, and Pádraig suggested we try to get past this, and get on with the interesting stuff.

To simplify things, we decided that it is probably best to use the grid partitioning system for considering co-location, as it gives us an implicit referencing scheme, and by considering neighboring locations to also be co-located, we can resolve the problem of split places.

With regards to the proximity calculations, we still need to verify the meaning of the signal strength, as is does not necessarily mean signal strength. We decided to consider any common WiFi point spotting to be a proximity event, and will record the start and end times of these, so that we can use metrics such as duration of contact in our routing scheme(s).

Davide offered to find out more about the signal strength issue, so that we can reliably refine it later on.

Pádraig suggested that  I/we need to find some good conference targets, ideally around January. I suggested ICUMT and Mobihoc, and Davide reminded me about a special issue journal call for papers that I spotted recently (Call for papers Phycom). Ubicomp 2011 is a possibility too.

By next week, I want to have the space partitioning method sorted out, and all contact/proximity events recorded. As well as coming up with an idea about the simulated on-device data structure, and some ideas about routing schemes.