Finally the thesis November 27, 2009Posted by tcarlyle in identity management, Me, Sensors, SIM Cards, Thesis, trust, Uncategorized.
Tags: context, identity, master thesis, sim
After almost six months that I have delivered my thesis, I’m finally posting it here. It turned up to be a very extensive document (about 150 pages), but mainly because we first wanted to assess the capabilities of SIM cards, identities and finally trust frameworks. And as I was working together with the SIM Research Team at Telenor and I do have some experience with SIM from when I worked in Gemalto, we spent several pages on reviewing the SIM capabilities and trying to figure it out the future SIM. We also touched an aspect that may start to become more present in the SIM cards which is the ability to sense context.
Other pages were spent in getting into the identity management world and this was one part of the thesis which in fact I wished I had more time to go through. I got very interested in going deeper in the field after finally understanding the identity frameworks such as Higgins, Cardspace and specially on the concepts in which they are based. At last we studied a bit about trust models and this was one of the most difficult parts of the thesis as none of us had much an idea of trust modeling and it is a topic that can get very complex if studied deeply.
After this long background, we finally chosen a new application that could be hosted in the state-of-art (or future) SIM cards, take advantage of the fact that the SIM represent one or more identities and that can be used to build trust. That application was what I have proposed in my paper mentioned in the previous post.
The idea is to use the future sim cards to sense each other (either through NFC, location information and server interaction, wlan, etc), to sense the environment and based on that, attribute a situational trust value for that meeting between the 2 sim holders. Then with a bunch of those situational trust value, you can infer the user relation. The more context information, the more you can infer.
Based on that idea, we made a small prototype using SunSpots representing those advanced SIM cards and with a simple trust inference model and a test scenario. It may sound a simple test and in fact it was, as the thesis focused a bit on bringing a new idea (which is extensively described) and the state-of-art research, having the prototype as a small proof-of-concept.
When I was reviweing the thesis for the paper presentation, I read in Bruce Schneier‘s blog about a paper from some researchers from the Santa Fe institute that used location information and phone calls information to infer the friendship closeness between the people involved in the experiment. The result was that they could predict the level of friendship with 95% accuracy! This pretty much confirm my thesis result =)