We must update our data structures to point out that all active drivers update their location every four seconds.
To overcome the above problem, we can use a hash table to store the latest position of the drivers and update our quadtree occasionally, say after 10–15 seconds.
Why not skip using hash-table and instead collect driver location every 15 seconds?
I was wondering the same thing as the usage of hash table is never explained. However, I found out that its purpose is to update broadcast the latest location of a driver to interested users. On the first getNearestDrivers() query, the clients will subscribe to the location of returned driver IDs (as in a pub/sub model). On each update, the location can be broadcasted using this hash table.
None of this is explained in the article . Very poorly written 