Global Connections

TerraStar PPP correction products are generated from the proprietary Orbit and Clock Determination System (OCDS). TerraStar operates three independently run network control centers, 2 in Aberdeen and 1 in Singapore, each with multiple, independent OCDSs. Operating multiple control stations and OCDS allows TerraStar to deliver the best possible availability, continuity, redundancy and quality of the service for its customers. To generate corrections, each OCDS uses data obtained from TerraStar's private receiver network. This network consists of more than 80 GPS and GLONASS reference stations located around the globe. These correction products are delivered to NovAtel CORRECT-enabled receivers via Inmarsat L-Band satellite communication links. By delivering corrections over satellite, machine guidance systems do not need local base-station infrastructure, cellular modem or Wi-Fi radio, greatly simplifying the user's hardware configuration. NovAtel CORRECT uses the correction data in advanced algorithms to provide highly stable kinematic positions with less than 5 cm horizontal RMS error.

PPP Error Mitigation

At the heart of all GNSS positioning solutions are range measurements and their associated observation equations. GNSS signals, however, are corrupted by a host of bias and other errors, such that the measured ranges can deviate substantially from the true ranges. Advancements in GNSS positioning have, to a large degree, been due to progresses made in the modelling and mitigation of these various error sources.

Error mitigation approaches can essentially be divided into three categories:

1. Signal combinations

2. Models

3. Externally-provided information

PPP is at the apex of GNSS error mitigation and uses all of the approaches listed above. For instance, to remove the effects of the ionosphere, PPP uses combinations of signals on different frequencies. Troposphere errors are reduced by troposphere delay models, and then further mitigated by zenith delay dynamic models. PPP correction providers supply corrections that remove the effects of satellite clock and orbit errors.

Research in troposphere delay demonstrates the advancements in error mitigation that have made PPP possible. When GNSS positioning was first calculated, the effects of the troposphere were disregarded altogether. Not long after, simple models using empirically-derived constants were introduced to mitigate the troposphere effects. These models were then refined and improved.

Later, it was recognized that the modelling error could itself be mitigated by estimating residual zenith delays in the receiver. The ability to correct the troposphere at this level makes PPP possible. Today, more sophisticated atmospheric models have been developed that incorporate troposphere-specific parameters in an attempt to reduce the troposphere error even further.

On the PPP corrections provider side, related marginal gains have made it possible to precisely estimate GNSS satellite positions and clocks in near real-time. This information is transmitted to a PPP client, like a NovAtel CORRECT with TerraStar enabled receiver, in the form of corrections to the broadcast orbits and clocks. There is some latency between the calculation of the satellite positions and clocks on the provider side and their use on the client side. Fortunately, however, the orbit and clock errors are well-behaved, and this latency can be accommodated by the PPP filter.

Ambiguity Estimation and Convergence

After error mitigation, the carrier-phase ranges from the GNSS satellites are effectively reduced to The net effect of the PPP error mitigation is to reduce the GNSS signal measurement precision to the amount of the remaining unmitigated errors. With a high-quality PPP correction feed, like TerraStar, this error is only a few centimetres. However, the ambiguity in the measurement still remains. The concept of carrier phase ambiguity is illustrated in Figure 1.

The figure shows how the receiver carrier phase very precisely measures a distance but it is not the distance to the satellite; rather, it is a distance to an unknown starting point. The distance from that unknown starting point to the satellite is the ambiguity.

Obviously, if the receiver antenna position was known, then the ambiguity could be instantly determined. Of course the receiver antenna co-ordinates are typically not known and the PPP filter must estimate them at the same time as it is estimating the ambiguities. This creates a circular dependency: the ambiguities are only improved by improved coordinates, but the coordinates are only improved by improved ambiguities. This coupling between the ambiguities and the coordinates takes time to resolve. This time is the convergence period.

One avenue for improving convergence is to improve the geometry of the solution, and the best way to do this is by adding additional satellites. Figure 2 illustrates how this improves the solution: the uncertainty in the ranges from the satellite leads to uncertainty in the position. This region of uncertainty is reduced as additional ranges are introduced. The TerraStar PPP feed includes corrections for both GPS and GLONASS, maximizing satellite availability and giving the best solution geometry.