Using cellphones for traffic mapping & planning


Stan Sieler
sieler@allegro.com
2015-01-30

What if iPhone and Android phones (and various other devices) had, 
in their operating system (not as a user application, like Waze
or any other), a bit of code that every two minutes woke up and 
reported location and physical speed (e.g., miles per hour) to 
a server somewhere.

I know that this isn't a new idea ... people have been experimenting
with it for years (at least since 2008).  But, my proposal is
to make it ubiquitous, efficient, and to try to preserve privacy.

The primary server(s) would pass the data to regional servers that
would produce aggregate information (e.g., Highway 85 northbound
at El Camino Real, congested...speed averaging 15 MPH), accessible
to major map providers (Google Maps, Apple Maps, etc.) and others
(just how many others, and the infrastructure to support massive
queries is a question for later).

What could we do with that data?

   1. Incredibly accurate and up-to-the minute accurate traffic maps,
      essentially world-wide (assuming cellular data service, of course).

   2. accurate data for planning streets/transit.

   3. accident / road jam detection, with the possibility of
      timely dispatch of emergency responders.

   4. remove need for costly in-road traffic sensors

   5. real-time traffic signal coordination

      (sitting at a red light with no cross traffic?
       complain! 
       ... this data and a smart signal would eliminate that!)

Why not an app?

   Apps that notice/report this information have a very bad
   track record in reducing battery life.

   Implementing a small, low-overhead mechanism as an operating
   system feature would result in (a) ubiquity; and (b) efficiency.

What about privacy?

   The server should not record incoming IP address, to alleviate 
   privacy concerns.
   (If the server(s) is compromised, privacy would be compromised.)

   Without an opt-in (mentioned later), privacy is pretty good 
   ... with the possible exception of a single vehicle travelling
   an isolated road (the scarcity of packets makes it possible
   to correlate them to the single vehicle on the road,
   which is a privacy problem).
 
   For the overwhelmingly vast majority of data submissions,
   there would be no privacy implications (because the data
   is anonymous, and not trackable back to a single vehicle).

   The software could limit pinging to whenever the device is in
   motion at, say, 20 MPH or more ... or had been within the last
   20 minutes.  Such motion would, for the most part, indicate the
   device is in a vehicle on a public road (hey, I said "for the most part!"),
   so since the device user is using a *public* road, they shouldn't
   complain about a tiny cost to improve the commonweal.

What about law enforcement?

   If the data shows that, say, 85% of the traffic on a given
   road is speeding, that could be used to revise the 
   speed limit upwards ... a practice already done by
   many states today.

What about data plans?

   The data packet should not be counted against data limits
   for phones / devices without unlimited data plans.

What about opt-ins to increase system usability for society?

   The first time a phone reports, it could generate a
   UUID and store it in a non-exportable area within the phone.
   The UUID would, with opt-in permission, be reported with the 
   location/speed information.

   Note: this feature has the most privacy implications ...
   even if there was no mechanism to export / view the UUID,
   one could correlate a UUID to a given phone, which would
   be a privacy problem.)

   But, associating a UUID with the data provides much better
   information for planners.  (E.g., what's the average trip
   length (distance and time)?)

What inspired this?

   Apple Maps this morning showed Highway 85 as "green".
   Google Maps correctly showed it as red, due to an accident.

   Had Apple Maps had access to accurate/timely average speed 
   information, I'd have gone a different route :)