Zinoviev Alexei

Chief Designer of the GNSS Unit
Alexei Zinoviev graduated from Moscow State University of Geodesy and Cartography (MIIGAiK) with a degree in Astro-Geodesy. After the graduation he worked for Russian Institute of Space Device Engineering, where he was involved in GLONASS issues. Then he worked for Moscow Rep. Office of Ashtech Inc. and Javad Positioning Systems Inc., developing firmware for high-precision GNSS receivers. Since 2001, he worked for Topcon Positioning Systems CIS, LLC, where he was in charge of the Department of GNSS Product Firmware Development. Since 2018, he has been working for NTLab-SK. He is SC104 Chair of IRNSS Working Group and SC-104 Chair of GLONASS Working Group RTCM committee.

Reports:

“SPHERE” CONGRESS SECTION Synergy: Using GLONASS as the Global Industry Technology for Creating Integrated Products and Services
«PPP Positioning Mode in NTLab GNSS Receivers»
Brief overview of Precise Point Positioning (PPP) mode. PPP advantages and disadvantages comparing to RTK. Information on GNSS measurement quality and positioning accuracy for NTLab GNSS receivers of geodesy accuracy class: these specifications meet the best world standards while maintaining the technological chain, under which all key components of GNSS receivers (digital and analog VLSI, circuity engineering, embedded software, etc.) are designed by NTLab, and thus do not contain any ‘black boxes’. This gives full control over components defining accuracy characteristics of GNSS devices.
Information on the methodology of PPP mode development. It is based on a conceptual model implemented using MATLAB/Octave. This model is convenient for implementing new algorithms and producing quick results of their evaluation. UFFI (Unified Format For Interfacing) files are used as initial data. UFFI is a unified binary format that can be used to convert the most popular GNSS data formats (RINEX, RTCM 3.X, CMR, etc.), including NTLab’s own binary format of transferring GNSS data. Such an approach is convenient in terms of the use of unified input data for modules based on MATLAB and C++.
At present, as a result of the work done, there are two PPP implementations: 1) MATLAB/Octave, 2) С++. The C++ implementation is fully portable to any software and hardware platforms and works in both post-processing and real time modes as a part of embedded software for NTL105 navigation module produced by NTLab. The set of algorithms is similar for both implementations and is planned to remain the same in future developments, using the MATLAB/Octave version as a means of testing new algorithms and approaches. Currently, the work with dual-frequency GLONASS/GPS measurements have been done, using two combinations of phase measurements: 1) ionosphere-free combination, 2) L1/L2 with ionospheric error assessment.
The accuracy characteristics of the implemented PPP mode were verified applying various initial data both in statics and kinematics. The PPP algorithms themselves were verified using measurements from many IGS network stations located in Europe, Canada, Australia, etc., which were supplemented by SP3/CLK files complying with final orbits and corrections from IGS clocks. According to these initial data, the convergence time to 10 cm horizontal error was between 2.5 and 16.5 minutes (on average, 8.5 minutes) and to 5 cm horizontal error – between 6 and 27 minutes (on average, 15.5 minutes).
Also, the similar results were obtained using an antenna installed on the roof of Technopark in Skolkovo and the stream of measurements from a NTL105 receiver. The diurnal files, processed by means of NRCan PPP service (Canada), were used to obtain the antenna’s precise coordinates: this solution’s accuracy in terms of latitude, longitude and altitude was around 1 cm. Using final orbits and IGS clock corrections, the PPP solution’s accuracy, comparing to NRCan coordinates, was around 2 cm in the horizontal plane, which was completely in accordance with results obtained using GNSS measurements from IGS network.
However, from the point of view of the PPP use, the main interest is on the performance in real-time mode. The public streams of PPP corrections (SSR messages in RTCM 3.X format), generated by various IGS organizations (CNES, CODE, GMV, etc.), were used to obtain PPP corrections in real-time mode. The conducted tests with the streams of real-time PPP corrections showed worse results comparing the use of IGS final orbits in post-processing mode, which was an expected result. However, 10 and 5 cm accuracy in real-time mode took around 15 and 30 minutes of convergence time. These results widely varied, which was also an expected result.
The tests in kinematics also demonstrated the high accuracy of implemented PPP algorithms. The horizontal accuracy with vehicle dynamics (50-100 km/h) was better than 5 cm in the horizontal plane (one sigma).
It can be concluded that the implemented PPP algorithms demonstrate the expected level of accuracy and convergence time and are ready to be implemented in various software and hardware platforms, supporting both post-processing and real-time modes. The future plans entail the expansion of systems and signals, with which PPP algorithms work (BeiDou, Galileo, L5, etc.), as well as the implementation of integer ambiguity resolution that should make PPP accuracy comparable to RTK accuracy.