Precise point positioning (PPP) using global navigation satellite systems (GNSS) enables accurate positioning worldwide. Recent advances, including improved error source modelling and the modernization of GNSS constellations, have reduced the time required to achieve cm-level accuracies from hours to seconds, creating new possibilities for several applications. To address remaining challenges in this area, the Remote Sensing journal has created a special issue dedicated to PPP.
In my last blog post, I talked about how Hexagon has tapped into multi-GNSS and multi-frequency processing to enable fast PPP convergence globally, without regional ionospheric corrections. To achieve this level of performance, attention to detail is crucial and I feel that the GNSS literature is lacking some of the information allowing researchers to replicate these results in a reliable manner. For this reason, I put together a special issue entitled “GNSS precise point positioning: towards global instantaneous cm-level accuracy”. The special issue’s official webpage is accessible here.
While many topics are relevant to the issue, here are some of the challenges that I would like to see being addressed:
- Time correlation: incorrect stochastic modelling in PPP typically leads to an overly optimistic covariance matrix. As a result, the reported uncertainties are often smaller than the actual positioning errors and ambiguity validation can become problematic. I wrote a blog post over 5 years ago on this topic, but the content is still relevant today. Can you make any of the proposed techniques work in PPP? Or can you come up with an alternative approach?
Ionospheric modelling: while it is possible to achieve fast PPP convergence without regional ionospheric augmentation, adding external constraints on the ionosphere will
still make convergence more rapid, consistent and reliable. Here are some research ideas related to the ionosphere:
- a one-to-one comparison (using the same underlying network) between the tomographic (3D) and conical (satellite-by-satellite) models. This blog post describes a similar comparison but was done using Kriging on the server-side only. Can anyone replicate this comparison using model parameters disseminated to users?
- Can a constellation of hundreds of LEO satellites contribute significantly to the tomographic model?
- Adding slant ionospheric delay constraints in a PPP filter can be detrimental to the solution if the uncertainties associated with the corrections are inadequate. I believe there is still some work to do to better characterize errors in ionospheric corrections. For example, if Kriging is being used, how is the covariance function adapting to changing ionospheric conditions?
- Ambiguity validation: I have been advocating BIE a lot as a useful ambiguity estimator for PPP. I also mentioned that BIE is a double-edged sword and can lead to wrong position estimates when the stochastic model is inadequate. How do you circumvent this issue? Can you propose another ambiguity estimator that works better in practice?
The deadline for submitting papers is 31 December 2021. Feel free to contact me if you need guidance or advice on your paper proposal. I can even provide a little bit of coaching if necessary. I hope that by focusing our efforts on these technical challenges, end-users in a wide range of applications including autonomous vehicles, marine navigation and many more, can experience fast and accurate positioning globally.