work is to 
[DAR and 
ould be to 
generation 
le imagery 
grammetry 
mmetry & 
entation of 
metry and 
ind control 
fication in 
unetry and 
er scanning 
nd Remote 
itching and 
1 for robust 
! Journal, 
adjustment 
grammetric 
apture and 
al Archives 
-388. 
nization of 
grammetry 
rithms for 
lata, /SPRS 
4(2-3):153- 
egistration 
orammetric 
ind Remote 
parameters 
cal Report 
Civil and 
OSU; 22 
R data and 
lescription, 
te Sensing, 
auspices of 
support of 
anbul 2004 
MULTISTATION ESTIMATING OF GPS SIGNAL ATMOSPHERIC DELAYS BY 
UNDIFFERENCED OBSERVATIONS 
f : b E 
A. Cina*, A. M. Manzino ', M. Roggero © 
Dept. of Georesource and Territory, Politecnico di Torino, C.so Duca Degli Abruzzi 24, 10129 Torino, IT 
* alberto.cina@polito.it 
t ambrogio.manzino@polito.it 
° roggero@atlantic.polito.it 
Commission I, WG 1/5 
KEY WORDS: GPS, Real-time, Algorithms, Software, Tracking. 
ABSTRACT: 
Precise RTK positioning is usually performable only on short length bases, to limit distance correlated biases, as atmosphere and 
ephemerides. However, it is possible to recover some residual biases also over long distance using appropriate area models, improving 
the performances of long base RTK positioning up to 100 km base length. Permanent reference station networks are used to estimate 
the bias model in surveying type applications requiring centimeter accuracy, reducing the worth of traditional single baseline methods. 
The well known advantages provided by reference station network information include improved modeling of the residual 
tropospheric, ionospheric and orbit biases. The system state vector, including the parameters of the bias model, is evaluated via 
Kalman filtering from the undifferenced observation equations. On the rover side, corrections are applied to the observations as single 
differences, then the model estimates and recovers the undifferentiated residual biases. 
1. MOTIVATION 
Continuous operating reference stations, tracking GPS or 
GLONASS, and in the future also Galileo, can be coordinated in 
static and also real time networks. Real time networks provide 
the support for surveying type applications, as for reference 
frame realization and maintenance, weather and space-weather 
forecasting, physical and geophysical type applications. The 
network properties can be defined under different aspects: 
e Provided products and services and their broadcasting 
infrastructure. 
e Network extension, density and distribution of the reference 
stations. 
e Network architecture, number and interconnection of the 
master stations, communication network. 
e Mathematical models and algorithms. 
e Integration of different source data, as meteo, radar, VLBI, 
laser ranging, etc. 
The paper focus on bias estimating and surveying type 
applications. To overcome the limitations of standard RTK 
systems, reference networks of permanent operating GPS 
stations can provide Network RTK (NRTK) services, 
broadcasting network based state parameters to the mobile user, 
who is then enabled to operate in Multi Reference Station or 
Virtual Reference Station mode. 
Network parameter broadcasting is limited in the RTCM 2x 
standards, but is now included in the new RTCM 3.0 standard. 
Standardisation discussion within RTCM in the new version 
design, has targeted the interoperability between reference station 
systems and rover receivers from various manufacturers. 
However, the obstacles in network parameters broadcasting with 
RTCM 2.x are overcame by the manufacturers by means of 
proprietary formats, sometimes largely accepted and used as 
FKP corrections included in RTCM message type 59, 
sometimes available in proprietary solutions only. The use by 
different manufacturers of the same proprietary formats, is due 
to the fact that interoperability is needed urgently. Moreover, the 
interoperability issues, are related to the creation and proper 
description of the models used for deriving the biases. 
The paper presents some remarks about NRTK techniques, 
taking account of results obtained by professional dual 
frequency and also low cost single frequency receivers. 
Data processing simulates NRTK procedures, using a self made 
software named ALARIS. This software performs one-way 
estimation of biases, real time static or kinematic positioning, by 
using different Kalman filtering techniques. 
2. DEFINITIONS 
In NRTK a master station estimates atmospheric biases over a 
network of permanent reference stations; then the rover station 
applies to his observations the corrections broadcasted by the 
master station. This positioning technique will be largely diffused 
in future for many applications in engineering, cartography, 
cadastre, GIS and so on. For all this applications, that usually 
don't require very high precisions, the use of low cost single 
frequency receivers is quick and economic (Cina, Manzino, 
Roggero, ISPRS VI/3, 2003). 
The proposed technique can be applied mainly in real time 
positioning and satellite navigation, but also in weather 
forecasting, ionospheric and tropospheric tomography on 
regional scale, time transfer, signal propagation and geomagnetic 
activity monitoring. 
To perform real time multistation estimating of GPS biases, the 
parameterisation in the state space domain is applied. For the 
state space approach the use of undifferenced observations is 
recommended, since this allows the estimation of clock 
parameters as well as the separation of atmospheric delays. It 
also makes the modelling of error processes easier, compared to 
. approaches based on double differences. However, using 
undifferenced observations we must correctly model some 
physical parameters, such as the corrections related to the 
receiver measurement style (cross-correlation or Z-tracking), 
satellite and receiver antenna phase centre, constant and periodic