The characterization of the adversary here being some malicious third party, though traditional, is something of a bogeyman as far as actual risk is concerned. Institutions running large mesh networks are in a privileged position to track users in this way.
At Brown University, complete connection logs for each wireless access point have been kept for no less than two years. A group of students last year were given this data and produced a (historical) map of student movement. There is no reason to believe that the University could not or is not rendering such a map in real-time.
As a general aside, when publishing documents that start off as LaTeX, putting them on Github (or some other Git hosting service) would potentially allow others to contribute corrections. There are a bunch of typos in the paper that would be trivial to correct, but of course there's no easy way to submit corrections to a PDF.
There is another post on my site about integrating live visualization with D3 as well.
I found that the metadata pertaining to signal strength to be too variable based on device manufacturer to be accurately used for a distance calculation without a lot of individual fine tuning or at the very least triangulation.
Triangulation runs into its problems though because of the inaccuracies of usb timing.
Interestingly, Apple has already started randomizing MAC addresses [1] on iOS (8?) and newer devices. Though maybe not quite as successfully as it seemed at first.
I believe this only works for broadcast packets. When you've chosen to connect to a wifi network, then your real mac address gets used. It would be a nightmare for shops that use mac filtering and other tools that expect standard mac behavior if the mac was randomized all the time. What Apple is doing is stopping broadcast packets being a unique value to track on.
At Brown University, complete connection logs for each wireless access point have been kept for no less than two years. A group of students last year were given this data and produced a (historical) map of student movement. There is no reason to believe that the University could not or is not rendering such a map in real-time.