Constraint Satisfier: We add all the users – their trajectories and
their home/work locations – in MongoDB database (version 2.6.7).
Then we measure the average query speed for the most expen-
sive operation, i.e. queries to the Trajectory collection – passen-
gers looking for drivers. Our results indicate that the average query
speed is around 2 seconds. Assuming a 32 parallel processes, then
we can answer 16 queries per second and we would need an hour
and 5 minutes for all the users in our database (assuming their
queries arrived in our system uniformly at random).
• Matching Algorithm: To evaluate the matching algorithm we do a
discrete time simulation where all the events appear in a simulated
timeline based on the order of their arrival. Assuming an ahead of
time notification of 20 minutes, our algorithm has a matching ration
of 78% when the offline upper bound is 80%.