Abstract: GNSS integrity assessment has always been linked to the need for reliable positional information. Initially used in aviation, positional information gained even more relevance in terrestrial applications with the popularity of GNSS. However, the terrestrial environment has many influences over GNSS signals, which reduces the positional quality of tracking objects. Advances have been achieved in the use of integrity monitoring algorithms, but there are limitations to their use, especially those concerning positional accuracy in urban environments with low-cost devices. This paper aims to discuss a comparative method using two low-cost GNSS receivers designed for transportation applications and to verify whether this method can evaluate positional quality in pre-established locations, as well as the possibilities of using these devices for transportation applications, considering the positional error. Results show that, in the static experiment, the receiver assembled with a GPS antenna active embedded was more accurate than the receiver assembled with an external antenna, presenting better values in 5 out of 10 evaluated sites, while the external antenna performed better in only 2 sites. However, in a kinematic evaluation, the receiver assembled with an external antenna provided better results when considering positional error as assessment criterion, resulting in values less than or equal to 8 meters in 99.7% of the route evaluated, while the embedded antenna had 95.3%.

Abstract: At the end of 2018, the Laboratorio de Meteorología Espacial, Atmósfera Terrestre, Geodesia, Geodinámica, Diseño de Instrumental y Astrometría, from La Plata University, started providing near real-time Regional Ionosphere Maps (MAGN) with high spatial and temporal resolutions and low latency. In this paper, we show a performance evaluation of these new ionosphere maps in space of positioning during the first year (2019) of available data. The assessment was carried out through absolute GNSS positioning using the in-house RT-PPP software. Data from five stations in Brazil, distributed in different geomagnetic latitudes, were processed. The performance of the MAGN was evaluated comparing the coordinates estimated using MAGN maps and the Global Ionosphere Maps (GIM) from CODE, both concerning the PPP ionosphere-free combination solution. Although it has been found that, in general, the MAGN and CODE solutions are compatible, the former provided better positioning result (8.5%) in the altimetric (up) component. However, when an atypical behavior in the MAGN solution at the end of the year 2019 as well as days of incomplete maps were removed from the analysis, the solution with MAGN becomes slightly better than CODE, providing an improvement of 8.6% in the up component.

To assess the impact of using gravity anomalies interpolated with Artificial Neural Networks (ANN) over geoid model compuation, this technique was used to interpolate Bouguer, free-air and Helmert gravity anomalies into a 5’ regular grid to compute the Santa Catarina State geoidal model. Three geoidal models were computed using the grids determined with ANN. The GEOIDE2014, computed by LTG (Geodesy and Topography Laboratory), was used for comparison. Therefore, the four geoidal models were assessed using information, provided by IBGE (Brazilian Institute of Geography and Statistics), of 53 leveling stations in which the geodetic altitude were determined with GNSS observations, called in this paper as GNSS/RN. The 53 geoidal heights were obtained from the difference between geodetic height and normal-orthometric height from the geometric (spirit) levelling. Such comparison has the aim of identify which models present better consistence with these points. It was obtained RMS of 0.17, 0.13, 0.17 and 0.17 m, where the geoidal model computed with the grid of Helmert anomaly interpolated with ANN presented the best result.

Este artículo trata de evaluar el impacto del uso de las anomalías de gravedad interpoladas con RNA (Redes Neuronales Artificiales), en el cálculo de los modelos geoidales. Por lo tanto, se utilizó de la RNA, para interpolar las anomalías de Bouguer, aire libre y Helmert sobre una malla regular de 5' con el objetivo de calcular el modelo geoidal de Santa Catarina. Se calcularon tres modelos geoidales utilizando las redes generadas con RNA. Se utilizó el modelo GEOID2014, calculado por el LTG (Laboratorio de Topografía y Geodesia) para la comparación. Los cuatro modelos geoidales fueron evaluados utilizando la información de 53 mediciones GNSS, sobre los puntos fijos de las líneas de nivelación fornecidas por el IBGE (Instituto Brasileiro de Geografía y Estadística). La comparación tuvo como objetivo identificar cuál de los modelos tiene mejor consistencia con la referida información. Se obtuvo los valores de los errores medios cuadráticos de 0,17, 0,13, 0,17 y 0,17 m. El modelo geoidal que utilizó la malla de la anomalía Helmert, interpolada con la RNA, mostró el mejor resultado.

Uma forma de se prever o conteúdo total de elétrons na direção vertical (VTEC - Vertical Total Electron Content) usando a arquitetura de redes neurais artificiais (RNA) denominada de perceptrons de múltiplas camadas (MLP - MultipLayer Percetrons) é apresentada e avaliada nesta pesquisa. As entradas do modelo foram definidas como sendo a posição dos pontos ionosféricos (IPP - Ionospheric Pierce Point) e o tempo universal (TU), enquanto que a saída é o VTEC. As variações sazonais e de períodos mais longos são levadas em conta através da atualização do treinamento diariamente. Testes foram conduzidos sobre uma área que abrange o Brasil e sua vizinhança considerando períodos de alta e baixa atividade solar. As RNA foram treinadas utilizando informações dos mapas globais da ionosfera (GIM - Global Ionospheric Maps) produzidos pelo serviço internacional do GNSS (IGS - International GNSS Service) das 72 horas anteriores à época de início da previsão. As RNA treinadas foram utilizadas para prever o VTEC por 72 horas (VTEC RNA). Os VTEC RNA foram comparados com os VTEC contidos nos GIM (VTEC GIM). A raiz do erro médio quadrático (RMS) da diferença entre o VTEC GIM e o VTEC RNA variou de 1,4 a 10,7 unidades de TEC (TECU). O erro relativo mostra que a RNA proposta foi capaz de prever o VTEC com 70 a 85% de acerto.

An approach to predict vertical total electron content (VTEC) using multiplayer percetrons (MLP) artificial neural network (ANN) is presented. The models inputs were defined as being the ionospheric pierce point (IPP) positioning and the universal time (UT), while the output is the VTEC. The seasonal and longer period variations of the ionosphere were taken into account by daily training the ANN. Tests were conducted over an area covering Brazil and its vicinity in both high and low solar activity. The ANNs were trained using the previous 72 hours of data from the Global Ionospheric Maps (GIM) produced by the International GNSS Service (IGS). The trained ANNs were used to predict the VTEC to 72 hours (VTEC RNA). Comparisons between VTEC RNA and the VTEC contained into GIM (VTEC GIM) were made. The root mean square (RMS) of the differences between them varied from 1.4 to 10.7 TEC units (TECU). The relative error shows the ANN can correctly predict the VTEC GIM from 70 to 85%.