The plan was to create the following issues in the WG tracker:
- create a registry for location obscuring algorithms
- describe the mechanism for algorithm selection
- describe the goals (not requirements!) for algorithms
- select/define and describe the old algorithm (this requires making the algorithm complete)
- complete and describe Jorge's algorithm
Only the first three of these are absolutely critical for us to progress the draft. We want to do that by May 15.
We have commitments from myself and Jorge to do this, but we'll need WG feedback if we are going to complete these tasks by our self-imposed deadline of May 15.
--
Some of the goals and constraints that we discussed (and may have agreed on to varying degrees) were:
Constraints
- The algorithm must take only one policy input: a distance in meters. Other parameters need to be fixed in the definition of the algorithm.
- The algorithm must describe what to do with uncertainty on the input location(s).
- The algorithm must be accurate [duckham05] - that is, the reported location must enclose the known location at the time that it is reported
Goals
- The algorithm should protect a static known location.
- The algorithm should protect the known location when the target moves. I admit that this needs to be more specific.
- The algorithm should protect the known location when the target returns to the same approximate location multiple times.
- The algorithm should limit the knowledge that an adversary gains about known locations at different times or locations, if an adversary learns of a known location for a given reported location. That is, if an adversary finds a known location, they can't use that information to (perfectly) derive known locations from other reported locations.
--Martin
_______________________________________________
Geopriv mailing list
Geopriv@ietf.org
https://www.ietf.org/mailman/listinfo/geopriv
