Network RTK and the Virtual Base
A single base station works well close to home, but its usefulness fades with distance, since the shared error that RTK depends on canceling stops being quite so shared once the rover strays far enough away. Network RTK solves that problem with an idea that surprises a lot of new users the first time they hear it explained clearly: the rover is often talking to a base station that does not physically exist anywhere.
A network RTK system uses many real reference stations spread across a region to model how atmospheric and orbital error changes with location, then generates a synthetic set of observations as if a base station were sitting exactly where the rover needs it, even though no such station is actually there.
Real Stations, Modeling an Imaginary One
A network of permanent reference stations continuously tracks satellites across a wide area. From those real observations, the network builds a model of how error, atmospheric delay especially, changes across that region. When a rover reports its approximate location, the network uses that model to generate what a base station at that exact spot would have observed, and streams that synthetic data to the rover as its reference observations. This approach is often called a Virtual Reference Station, or VRS.
A virtual reference station is not a physical monument anywhere. It is a modeled estimate of what a base station would observe at the rover's approximate location, generated from a real network of surrounding stations.
Why Go To the Trouble
The benefit is significant: a rover working with a good network correction gets a baseline that behaves as if it always has a base station nearby, no matter where in the network's coverage area it happens to be working, without anyone hauling a physical base station out to every job site. The trade-off is that this convenience depends entirely on network coverage, connectivity, and the quality of the underlying reference station network. Step outside network coverage, and that virtual convenience disappears.
A network correction is not simply a more powerful version of a single base station's signal. It is a fundamentally different service, generating a synthetic set of observations from a modeled network rather than broadcasting from one specific, physical point.
The Rover's Job Has Not Changed
Everything from Chapter 3 still holds here. The rover still solves its own position independently, differencing the observations it receives, whether those observations came from one physical base or an entire modeled network, purely to cancel shared error. The network does not compute the rover's position any more than a single base station ever did.
Network RTK generates a synthetic set of base station observations from a modeled network of real reference stations, delivering them as a virtual reference station tailored to the rover's approximate location. The rover receives and processes these simulated observations exactly like a standard physical broadcast.
That correction data, wherever it comes from, still has to travel from the network to the rover somehow. The next chapter looks at the protocol that carries it, and why understanding that protocol prevents a very specific kind of confusion.