rtk-rs / gnss-rtk

Precise Positioning in Rust

GNSS-RTK

Precise positioning calculations, in Rust

Notes on this ecosystem

• Nalgebra is used for anything related to linear algebra
• Nyx-space provides required features in orbital calculations
• Hifitime provides timing and timescale definitions
• Gnss-rs provides basic GNSS definitions

The RINEX Wiki describes extensive application of this framework (at a high level).

Solver

We only support 1D direct positioning, a subsidary data source required for RTK is not supported and will be developed very soon.
That means the only differential technique you can currently used is SBAS (Ground/Space differential technique).

PVT Solutions

The objective is to resolve PVT solutions, which requires a minimum of SV in sights:

• 4 SV required, in default mode
• 3 SV required, in fixed altitude mode
• 1 SV required, in time only mode

The preset criteria are manually set in the configuration file (or config script). At the moment, refer to the RINEX Wiki or RINEX scripts database, for meaningful examples.

Depending on the preset configuration, other requirements will apply to the previous list, most importantly:

• CPP strategy will required pseudo range observation on a secondary frequency
• PPP strategy will required pseudo range and phase observations on two frequencies
• SNR, Elevation and Azimuth mask will require to gather the required amount of SV within those conditions

Each PVT solution contains the Dilution of Precision (DOP) and other meaningful information, like which SV contributed to the solution. We have the capability to express the clock offset in all supported Timescale.

Strategy and other settings

The solver's behavior and output results are highly dependent on the selected strategy.

Advanced strategies require deeper knowledge and most likely more tuning of the solver configuration. The Rust/JSON infrastructure is powerful enough though, to allow to only define the config parts you are interested in: others will simply default.

PVTSolutionType defines the type of solutions we want to form and therefore, the minimum amount of SV we need to gather. As previously stated, other criteria like min_sv_elev or max_sv_azim will restrict the condition on those vehicles that they must fit in to be considered.

When fixed_altitude is set to a certain value, the quantity of required SV is reduced by 1.
This has no impact when PVTSolutionType is set to TimeOnly.

The SolverOpts configuration gives more advanced options on how to tweak the solver. Briefly, this allows to

• select one of our Navigation Filters, like Kalman filter or LSQ
• define the PVT solutions confirmation criteria

Modeling defines what physical and environmental phenomena we compensate for.
Modeling are closely tied to the selected solver strategy. For example, models that impact at the centimetric level like the sunlight rate, are not meaningful in strategies other than advanced PPP. On the other hand, you will not reach metric solutions, whatever the strategy might be, if a minimum of physical phenomena are not accounted for.

Atmosphere and Environmental biases

This solver is always capable of modeling all conditions and form a solution.
It is important to understand how our API is designed and operate it the best you can to get the best results.

Troposphere Delay

Troposphere bias always needs to be estimated.
By default, the solver will use a model implemented in the [model::tropo API].
If you're in position to determine yourself the Tropospherical Delay components (TropoComponents structure) at the required latitude and Epoch, you are highly encouraged to provide your data. To do so, we use a function pointer that can serve as a TropoComponents source. TropoComponents evaluation parameters (function pointer arguments) should be :

• Epoch
• altitude (above sea level) expressed in meters
• latitude expressed as decimal degrees

For Epochs where the data source is not capable to supply data, that is not a problem, we will rely on the internal model.

Example of handmade TropoComponents provider :

TODO

⚠️ When using the internal model, we recommend applying a >= 5° elevation mask to the candidates.

Ionosphere Delay

TODO

A priori position

The solver can be armed with a priori knowledge (rough idea of the final position), or can operate in complete autonomy. In this case, the solver will initialize itself very accurately, this requires one extra step.

Getting started

Refer to our example applications to understand how to operate our API in more detail.