Towards Airborne Multipath Models for Dual Constellation and Dual-frequency GNSS

Abstract

This paper presents the initial results of an effort to develop and establish airborne multipath models for the GPS and Galileo L1/E1 and L5/E5a signals. These models will be used in dual frequency and multi constellation aeronautical navigation techniques, such as ARAIM, SBAS and GBAS. Development of the models is based on electromagnetic simulations and a large-scale data collection campaign on a variety of different airframes. In this work, results of the evaluation of the first batch of about 200 hours of flight test data with experimental equipment are shown. At a later stage, more data with avionics hardware will be collected and evaluated. This paper mainly aims at establishing the methodology and obtaining initial results as a first indication. For this purpose flight data is analyzed for periods where satellites are tracked continuously. Then, the carrier phase measurements are subtracted from the code measurements, the ionospheric divergence is removed and the bias introduced by the carrier phase ambiguities is removed. Finally, all the data is sorted, overbounded and inflated to account for the limited number of samples. The methodology also includes the removal of code errors induced by the antenna group delay, as these are suggested to be treated separately from the multipath errors.The absolute values of the bounds for GPS L1 and Galileo E1 are larger than the existing GPS L1 model. This could be expected when being conservative at each step of the evaluation process and when using experimental hardware that is not entirely compliant with current minimum performance standards for airborne equipment. Without the conservatisms applied, the observed errors are below the Airborne Accuracy Designator (AAD A) curve. The level of residual multipath on L5/E5a is significantly smaller than that on L1/E1 and the model for the ionosphere free combination is significantly larger than the bound on L1/E1. Both these properties found in our experimental results match with theoretical expectation.

Initial Results for Dual Constellation Dual-frequency Multipath Models

“…6. Overbounding of the data to safely bound the tail risk as described in [4]. This step is applied independently to each frequency and the ionosphericfree combination of the signals of the two frequencies.…”

“…In an initial phase towards developing the airborne multipath models around 200 hours of experimental flight data collected with an experimental hardware installed on DLR's A320 research aircraft was analyzed. The methodology was presented in detail in [3] and preliminary results collected with this experimental installation were shown in [4]. The previously derived versions of the models were more conservative and exceeded the existing 𝜎 𝑎𝑖𝑟 models for GPS L1.…”

Standardization of New Airborne Multipath Models

“…The methodology for the development of the new dual frequency GPS and Galileo airborne multipath models was presented and discussed in [7] and [8] and presented to the ICAO NSP, RTCA SC-159 and Eurocae WG-62. The methodology consists of two main branches: a simulation-based approach and a measurement-based approach with a matching of both methods.…”

“…As the number of available samples is small that step is omitted for the time being. The methodology and all the above shortly summarized steps were described in detail in [8]. Finally, a model describing the standard deviation of the multipath as a function of elevation is derived.…”

Validation of the GPS leveling method through the gradient analysis of the geoidal wave. Case study of Ecuador.

“…The first consisted of satellite tracking the base leveling plate using the relative static method, with an approximate time of 8 hours, using a dual-frequency GNSS equipment that guarantees the link to the GNSS Network for Continuous Monitoring of Ecuador (Yunck et al, 1985;Marais et al, 2005;Dow et al, 2009). In the second phase, the other leveling plates were determined through the fast static method, with a tracking time of 1 hour, also using dual-frequency GNSS equipment (Felux et al, 2019;Robustelli et al, 2019).…”