Our primary emphasis has been on modeling and estimation of multipath errors in GPS carrier phase data for the purpose of improving the accuracy of GPS for attitude determination and precision orbit determination. Significant contributions were made by former students Dr. C.J. Comp and Dr. A.K. Reichert in the use of measured signal-to-noise ratio as a means for identification and correction of multipath errors in data. Dr. Comp’s work centered on adaptive techniques for estimation of multipath in changing environments. Dr. Reichert focused on identification of relatively stable sources of multipath for static ground environments and fixed reflectors in a spacecraft environment. We have also applied a phase error mapping technique originally proposed by Cohen and Parkinson of Stanford University to a dynamic spacecraft environment. To address the contribution of carrier phase multipath errors to the error budget for precision orbit determination on missions such as Geosat Follow On (GFO) and IceSAT, we have used the Ohio State Basic Scattering code for prediction of multipath errors and worked with orbit determination codes such as JPL’s Gipsy Oasis-II (GOA-II), and Microcosm. We are currently looking to extend these methodologies to other applications.

The vast majority of my work has dealt with multipath characterization rather than multipath mitigation. However, I have developed a concept for a multipath-mitigating antenna for DGPS ground reference stations which will be the standard antenna for the United States Local Area Augmentation System (LAAS). In addition, I have studied most of the current GPSreceiver architectures including those which claim to have multipath-mitigation properties. These include Narrow-correlator, MET, MEDLL, Strobe, Enhanced-Strobe, etc.

Our current work in multipath analysis is still concentrated on the use of the signal to noise ratio technique for estimating the amount of phase multipath on individual satellites. The main input to the technique is antenna gain pattern information and the signal to noise ratio itself. We have had considerable success with certain kinds of equipment (where high fidelity SNR data is available) but only moderate success with other receivers. In general, we can get a positioning improvement in the range of 20%-50%. We are also working on multipath mitigation through stochastic modelling. In this case we are using output from the phase smoothing process to develop a weighting strategy for phased smooth pseudo-ranges when operating in DGPS mode. We have not yet published any results from this latter activity.

At the moment I am trying to put together a project that will try to understand multipath better in predominantly static environments so that very small deformations can be detected. It is likely that this work will begin towards the end of this year.

One of the main research activities of our group (Department Engineering Sureveying and Metrology / TU Graz) focuses on quality control of GPS signals. We have developed weight models (SIGMA models) for treatment of signal attenuation and signal distortion. These models use the carrier-to-noise power ratio to weight GPS phase observations according to their quality. Using these models the accuracy of GPS results is improved by more than 50% compared to standard GPS processing techniques.

Beside this work we investigated antenna characteristics in high and low multipath environment. These tests indicate that choke ring antennas reduce multipath effects by a factor of about two compared to standard geodetic antennas and that choke ring antennas are insensitive to ground multipath. However, even if a choke ring antenna is used in low multipath environment (no vertical reflectors in a circumference of several km) significant deterioration of the phase measurements exists. This effect will be further investigated.

Recently, a semi-parametric model using the penalised least squares method has been developed for mitigation of multipath errors at Curtin University. In the semi-parametric model, multipath is described by a complicated, but smoothly varying function with time. The function parameters, such as the coordinates of sites and ambiguities, and observation noise are decomposed using the penalised least squares method. Several data sets, which were collected from different observation environments including an open pit mine, a dam and a site surrounded by trees, have been processed using the developed method and software. The results have shown that the semi-parametric model can reduce multipath to the almost level of receiver noise and that ambiguities can be fixed successfully using much shorter date sets compared with most commercial software packages. Baseline length accuracy has also been demonstrated to improve using the semi-parametric model.

At moment we are developing a statistical test procedure detect if there exist significant multipath errors in a specific data sets and then decide if it is necessary to act a backfitting routine after least squares solution in order to save time in the solution procedure without reducing accuracy and reliability of the results. In order to compare different multipath mitigation technique we will also try in next few months to process the same data sets using some existing methods such as a variety of weighting algorithms.

UNICAMP, Brasil

Current work of our research group deals with mitigation of multipath effects by means of adaptive signal processing techniques, such as smart antennas, neural networks and unsupervised filtering techniques (blind equalizers). Satellite telecommunications and GPS are the major applications. We have also implemented a GPS simulator including models of multipath effects. The relationship between solar flares and multipath effects are currently understudy, with the purpose to include these phenomena in the simulator, as well as to develop particular mitigation techniques for special solar events.

I am mainly working on multipath mitigation techniques for GPS-based attitude determination systems. In particular my work focus on the definition and testing of software algorithms for multipath compensation and the improvement of the system performance through design of specific hardware components such as antennas. These activities are carried out in the framework of R&D contracts with

European companies and institutes and as internal R&D studies. Internal studies are supported by a GPS test facility consisting of a tilt table, a GPS GEC-Plessey receiver modified to receive differential phase measurements and various sets of antennas. The facility is used to validate algorithms and assess the system performance.

At present we are involved in the following collaborations: with University of Surrey for the UoSAT12 mission that is flying a GPS receiver as an attitude and navigation sensor; with Matra Marconi Space for the development of software algorithms; with University of Calabria for the design of optimised antennas for multipath rejection.

My present work focus in multipath mitigation techniques at software level. We are currently implementing a new multipath mitigation formula in an European space qualified 12 channels NAVSTAR compatible receiver.

I am also heavily involved in multipath detection and mitigation on the ESA´s Automatic Transfer Vehicle (ATV) (http://www.estec.esa.nl/spaceflight/) which will reboost and re-supply the International Space Station (ISS). The RendezVous and docking operations of ATV will be carried out using relative GPS.

Multipath Characterization  GPS receiver code and carrier tracking loop discriminator function responses are analyzed in the presence of multipath signals. The characteristics of multipath errors are shown in terms of the mean, standard deviation, pattern and overall envelope through simulations for various types of discriminator functions. The influences of the antenna-satellite and antenna-reflector geometry on multipath frequency are analyzed by forming relationships among them. The spatial correlation property of multipath within a small area are investigated by relating the antenna-satellite and antenna-reflector geometry using multipath and geometric parameters, such as, the reflection coefficient, multipath delay, multipath phase, as well as multipath signal azimuth and elevation. 

Multipath Mitigation Using Multiple Antennas An algorithm to mitigate code and carrier multipath errors using a multi-antenna system is developed. A Kalman filter is developed to use multipath-corrupted measurements from multiple closely-spaced antennas to estimate the multipath and geometric parameters, from which the multipath errors in the code and carrier measurements at each antenna can be computed. Field tests in a moderate multipath environment show a reduction in multipath errors up to 73% (average 22%) in the code and up to 52% (average 15%) in the carrier residuals. Improvements are also observed in the position domain, whereby the differential position accuracy is improved by up to 51% (average 21%) and up to 37% (average 24%) for the non-smoothed code and carrier cases respectively. A desirable characteristic of this technique is that it is more effective in a high multipath environment. This technique has potential to be used in real time in reference stations generating corrections for kinematic applications. Furthermore, this technique can be extended to other direct sequence spread spectrum communication systems employing PRN codes for signal spreading.

In 1992 I had a grant (Telespazio s.p.a.) working for the Politecnico of Milan, to study the error budget of the GPS signal. Currently I am particularly interested in multipath problems in high precision surveying and in automatic vehicle navigation. On the subject of vehicle navigation, my collaborator ing. Gianluca Fruet is has a grant in cooperation with the Centro Ricerche Fiat until December.2000. 

My research topics are:

- critical satellites configuration for multipath effects,

- to avoid positioning multipath error in land mobile navigation

- (C/A observation) from the 3D obstacles Digital Surface Model;

- real time algorithm to detect multipath effects.

Rod Walker

Queensland University of Technology, Australia

This year has seen Dr Feng and myself switch our efforts full-time to the applications of GPS in the space environment. This is in keeping with the CRC for Satellite Systems such  we are both participants in. One of our research programmes is in the multipath mitigation of GPS for use in precise orbit determination applications on small spacecraft. JPL's GPSMET programme was consistantly capable of achieving an orbit accuracy of approx. 5cm (cross-track error) and approximately half of this error was attributed to multipath from the spacecraft structures. We have spent the last year modifying our EM propagation models to work over small domain sizes (of few wavelengths) to allow us to compute propagation profiles over the spacecraft (as opposed to the large mines that we had previously analysed). We have also spent a substantial amount of time developing new receiver models such that, based on a particular correlator and discrimination function design, we can predict the range error. This modelling allows us to predict the amplitude, phase and phase rate of change for the signal received at the antenna and then using our receiver model we can predict the magnitude of the range error for a number of different receiver designs. We have done a complete analysis of FedSat (the CRCSS satellite) and found that multipath will not affect significantly our aims for that mission.

We have also recently purchased a $70K Welnavigate GPS constellation simulator. This machine completely simulates the RF environment of the GPS constellation. We plan on using our EM models as inputs to the settings on the constellation simulator and to characterise the multipath out of a real GPS receiver. We have been working with SigTec Pty Ltd (manufacturer of GPS receivers) and they have provided for us a developmental GPS receiver that we can modify the software in. This new equipment gives us the means to take the results of our modelling and start applying it practically.

Future developments are to include antenna gain profiles in our EM models. We have done these analysis in the past, however the method of modelling was not rigorous. We now have a technique for modelling antenna types much more accurately.