Using only the connection itself, that is, the cell towers, one can track a phone to within a few hundred meters or better depending on where they are.
There is no need for the phone to send a file, they are constantly sending signals to nearby towers as part of their "presence" systems. This is how the network knows where to send incoming connections like calls and texts, it sends it to the tower with the strongest signal.
The system works by sending signals which the tower responds to. By timing the round-trip time it draws a sphere of that size around the known location of the tower. By producing several such spheres, the phone is located by the surfaces where they intersect. If you think about it in 2D you have two points where two circles intersect, and some other system needs to be used to pick which one you're at. If you have a third tower you can remove that ambiguity.
Since the surfaces are 3D and the towers know their location very accurately, you can recover altitude too. This has been done on drones carrying cell phones with no GPS signal. The accuracy was enough to avoid hitting the ground in height, but not enough to avoid trees or buildings in X-Y.
The quality of the tracking depends on the density of towers. So its really very good indeed in cities and on roads, but gets progressively less accurate as you leave towns. In my case, at the cottage there's only two visible towers so I get results within about 2 km. So for the movie, it's very realistic to have someone being tracked in this fashion - it's how E911 works for instance.
This system was first used in WWII. The Germans had a system called "Wotan" that measured range from one station and angle from another, so it wasn't true multilateration. The RAF introduced their "Oboe" in 1941 which used two transponders to produce two arcs with the target at the intersection. So it's not a new concept! GPS works exactly the same way mathematically, but only uses one-way signals. To make this work you need to know when they sent it so you can time the delay, and they do that by sending the clock in the signal. Very clever.
So for a movie, let's say someone put a cell phone on a drone to use its camera (which is already common) then you could probably track that in a city to under a block. And you could even figure out it was a drone by measuring altitude, but only in dense areas - perhaps a dialog point when it flies into an open area, perhaps an intersection, and the quality improves and they notice it's not on the ground.