Grasso Toro, F.; Diaz Fuentes, D. E.; Schnieder, E.:
Basic Intelligent Models for Validation of Dynamic GNSS Measurements.
ADM 2014 - 8th Workshop on Analysis of Dynamic Measurements, Turin, Italien, May 2014.


Localisation systems based on Global Navigation Satellite System (GNSS) need an evaluation of the generated position information related to an independent reference. This paper describes a method for GNSS receivers’ validation by means of an extended accuracy evaluation of dynamic GNSS data, implementing an intelligent accuracy-based quality function (iAQF). The proposed iAQF consists on an artificial neural network (ANN) combining quantifiable measurements from GNSS receivers’ data (inputs) and postprocessed accuracy-based deviation analysis results (targets). The developed artificial intelligent (AI) system is trained to describe the measurement uncertainty by means of trueness and precision. In the ANN the results from a Mahalanobis Ellipses Filter (MEF) accuracy-based deviation analysis are used as targets, while the inputs consist on several GNSS measured variables, such as the number of satellites in view, their signal to noise ratio (SNR), the horizontal dilution of precision (HDOP), etc. The obtained measurement uncertainty can be used in further works to estimate the quality of the dynamic localisation system in real time scenarios. For this purpose three dynamic scenarios are proposed for the validation of the iAQF: 1) open area roads, 2) normal city roads and 3) narrow city roads. While the first scenario considers dynamic measurements performed at constant speed, the second and third scenarios consider dynamic measurements performed at variable speed. For one specific scenario actual results obtained by testing a developed ANN model are shown and a brief description is given. The AI-based validation approach for dynamic GNSS data will allow further development on intelligent accuracy-based quality functions and certification processes for GNSS receivers.