So, when a base station is located close to the rover, RTK eliminates all shared errors between the rover and base, allowing maximum use of phase measurements.

When the rover gets the corrections, it just "simply" (of course, not very simply) applies them to its own measurements and gets the desired precision. And it’s the best we can currently achieve in GNSS positioning.

But that raises a natural question:
— Why have other positioning methods emerged and evolved?

The answer lies outside the realm of accuracy and convergence time:
RTK requires dense and expensive infrastructure — a network of base stations. Network RTK attempts to reduce the requirements for net density, but networks still need to be rather dense. All other methods (PPP / PPP-RTK) are essentially further attempts to reduce the infrastructure burden.

This fact also stimulates the development of methods like PPP / PPP-RTK, which can use unidirectional broadcast communication channels.

In RTK the rover should get corrections only from the nearest RTK station to achieve precise location. And if you have hundreds or thousands of stations in your RTK network, you can’t broadcast the corrections of all stations to all rovers because of bandwidth limitations. That is why RTK requires a non-broadcast bi-directional communication channel, which is rare to find in a crop field, or requires some hacks to provide only the required corrections in a certain area.

?

Section 02

However, because the network is sparse, atmospheric corrections are not computed, and we can’t reconstruct the full measurement at the rover level — so we can’t use a simple RTK-like approach, which naturally imposes limitations on the time to first fix.

Fig 02

Before we go further, let’s briefly review what PPP — Precise Point Positioning — actually is. In PPP, you need orders of magnitude fewer ground stations to collect raw GNSS signals and calculate satellite orbits and clocks.

Once the satellite orbits and clocks are calculated, this data must be delivered to the rover. After receiving it, the rover starts processing its own measurements with PPP corrections provided by the computational service, along with the GNSS signals it observes from the sky.

We get a solution with a sparse base station network and small bitrates that can be broadcasted — but with a long fix time.

This brings us to the next question:
— Сan we achieve high precision with a sparse base station network, but with faster convergence?

PPP / PPP-AR

Section 03

Fig 03

Such attempt is PPP-RTK. It’s designed to take the best from both positioning worlds: absolute (PPP) and relative (RTK).

On one hand, it uses precise orbits, clocks, and biases — just like PPP — allowing it to achieve comparable performance. On the other hand, instead of "reading" tropospheric and ionospheric error directly from a base station, it leverages tropospheric and ionospheric information, computed during processing from the base station network, to model the measurements and operate in a pseudo-RTK mode.

This gives the rover fast convergence time and high accuracy, even in any area with at least partial access to a base station network. PPP-RTK is significantly more tolerant of base station distance and comm link interruptions compared to RTK. In RTK, these parameters are critical — their degradation reduces the percentage of fixed solutions.

In PPP-RTK, base station density and communication channel bandwidth mainly affect only the convergence time (if a fix hasn’t been reached yet).

PPP-RTK

PPP-RTK

Valuable advantage of PPP-RTK is lower correction transmission requirements. Unlike RTK and VRS, the corrections can now be broadcasted, including via satellite. This is made possible by the SSR (State Space Representation) approach, which transmits model parameters instead of raw base station observations (OSR - Observation Space Representation).

Building on this, PPP-RTK technology enables the generation of a Virtual Reference Station (VRS) on the client side using SSR data - essentially allowing legacy RTK-only receivers to receive corrections via a broadcast channel.

This technology is known as SSR-to-OSR, and is already widely used in markets like automotive and IoT. Considering the high degree of customization in the interaction between the rover and the provider, such an SSR-to-OSR adapter usually is part of the provider’s software stack.

Finally, just like PPP, PPP-RTK provides much longer standalone precise positioning after signal loss compared to RTK: tens of minutes vs. just a few. This is possible because it uses a mathematical modeling approach instead of computing corrections based on direct observation of GNSS errors.

PPP-RTK and SSR with SSR2OSR