Establishment & Testing of Dubai Virtual Reference System (DVRS) National GPS-RTK Network



2. GPS Network
The concept of GPS Network Reference Stations allows us to eliminate/reduce systematic errors in reference station data, i.e. allows to increase the distance to the reference station for RTK positioning while increasing the reliability of the system and reducing the initialization time. The GPS Network Reference Stations requires continuous modem line connections between the control center and all reference stations. Data is transmitted continuously to the center. The center will calculate and transmit optimized Radio Technical Commission for Maritime Services (RTCM) correction messages and transmits it to the users.

Permanent GPS reference stations making it possible to receive RTK (Real Time Kinematics) correctional data everywhere. This means centimeter precise measurements in real time. By establishing an adequate number of GPS stations and allocating data access to the reference stations via mobile phone using Global System for Mobile phones (GSM) data modems, they can offer correction data throughout the region of interest. All the GPS reference stations in a network send "on-line" raw GPS data via permanent connections to a super-computer housed in a secure Control Center. In this way all GPS observations can be gathered and weighted to the user's advantage. This solution gives the following advantages:
  • Uniform precision of the entire network, or in other words, no additional constants due to increased distance from the individual reference stations (a well-known problem in traditional GPS RTK surveying).
  • Single correction data from the entire network.
  • Safety and reliability to enhance the quality of GPS measurements.
The Control Center of super computer takes care of the following numerous tasks:
  • Import of raw data and quality assurance routines
  • Storing of RINEX data
  • Correction of antenna phase center wandering
  • Modeling and estimation of systematic errors
  • Calculation of correction data in RTCM format for the users
  • Transmission of data to users in the field
wss There exist many Networking approaches where GPS signals corrections can be send to mobile rover in the field for Real Time Psositioning.

2.1 Virtual Reference Station
An RTK rover located near the center of several reference stations would be affected by systematic errors if using any one of these reference stations. If, however, measurements from all these reference stations are combined, a model of the geometric and atmospheric errors over the area can be determined and a VRS can be created adjacent to the rover's location, dramatically reducing the systematic errors.

2.1.1 Virtual Reference Station (VRS) Technique
A pre-requisite of the virtual reference station (VRS) concept is the need of a duplex communication link between a node of the reference station network and the rover. The rover has to transmit its approximate coordinates to the network, which then interpolates from the state information a reference data stream VRS k/v for the given position. The data relates to the observation space ref Wübbena et. al (1996):


Equation (4.20) contains a tropospheric term , which describes the difference between the tropospheric delay models used in the network processing on the original reference station and the virtual reference station. Due to the RTCM definitions, the reference station may not correct for tropospheric errors. This is in general a reasonable restriction, because it avoids the problem of using inconsistent models for reference station and rover, while the rover is responsible to compute corrections for both sides. This, however, requires the knowledge of the reference station coordinates at the rover. Since the only coordinates the rover knows about are originating from the RTCM data stream, the rover does only know the coordinates of the VRS. Hence, the rover cannot compute the tropospheric correction for the real, but only for the VRS. In consequence, the network has to apply the tropospheric correction between real and virtual reference station. And here there is again the problem of possible inconsistency, if it is done with a different model than the rover applies.

In the VRS concept, the coordinates (in RTCM message type 3) are changed to VRS location, hiding the true reference station completely from the rover. One disadvantage of the VRS concept is, that for a kinematic rover continuously updated approximate coordinates have to be used for the VRS computation (moving reference station). Today, most rover systems cannot handle a kinematic reference station. A system reset is performed, if the VRS coordinates are changing, which will result in frequent initialization of ambiguities. In practice, the VRS position therefore does not change. However, this implies that distance dependent errors will be present in the rovers solution once it starts to move away from the virtual reference.

Typically, some irregular physical effects occur, which can hardly be determined by a reference station network with given station distances. In this context, the reference station network can be considered as a limited number of monitoring stations or sensors with a certain and restricted spatial capability. The errors may arise from local troposphere or turbulent ionospheric conditions. Even if these higher order errors cannot be determined by the reference station network, it is obvious that their magnitude is a function of distance from the next true reference station.

Thus, if the rover knows the reference station position(s), it can take into account these higher order errors and improve its own RTK models, e.g. by stochastic ionospheric modeling. If the rover knows only the VRS position, it has no chance to do such kind of improvement. It should be mentioned that there are different types of VRS depending on the type of networking model. A VRS derived from the observation space (OSP-VRS) shows different behavior than a VRS derived from a state space model (SSP-VRS). This results from the fact, that a SSP-VRS is much less affected by current individual reference station errors than the OSP-VRS ref. Wübbena 2001.

Since the state vector is the result of a continuously running filter, the influence of station dependent errors reduces, the more redundancy (number of stations and satellites) is available in the network. A similar filtering in the observation space can only be done with arbitrary models and is therefore less effective. Especially the non- depressive part of the signal is much smoother if derived from state information than from the observation state.

2.2 Area Corrections Parameters (FKP) Technique
One way of representing the additional corrections for the distance dependent errors is a polynomial parameterization to describe the influence for any rover position in a certain area. Depending on the temporal and spatial variation the orders of the representation must be defined. The RTCM standard currently limits the correction data to be formulated in the observation space, which means, that modified GPS observable must be used. The area correction parameters (commonly called FKP), are the most flexible and suitable way to represent the state. FKP can be assumed for this discussion as a representation of the full state space information. FKP are more or less simplified to reduce the required bandwidth for transmission or the complexity to apply it at the rover. The state has to be transferred to the observation space, because most rover systems are currently not capable to handle any state space information. The FKP allow the prediction of the distance dependent error term for the approximately known rover position:


This can be done independently from the network processing, as only the rover coordinates and satellite information are required. It is a major advantage, that FKP can be distributed by broadcast media, which is requested by most service providers. The FKP do not contain absolute tropospheric information, but gradients of the troposphere. The tropospheric effect for a reference station can therefore be figured out and applied correctly to the data by the rover.

The dimensions of networks and the coverage of distribution media often make a linear FKP representation sufficient. The coverage of a linear FKP model is

then centered to a real reference station, and the FKP describe the horizontal gradients for the geometric and ionospheric signal components in the observation space (Figure 2.2).


Figure 2.2 : Linear FKP planes for four reference stations

3. Problem Statement
There are multiple objectives for the establishments of Dubai Virtual GPS Reference System(DVRS) network, such as follows:
  • Real Time Kinematics (RTK) applications
  • Differential Global Positioning System (DGPS) applications
  • Realisation and continuous improvement of the International Terrestrial Reference
  • Frame (ITRF)
  • Realisation and continuous improvement of the Dubai Emirate Special Reference
  • Frame
  • Absolute sea level determination
  • Monitoring of the deformation of the earth
  • Facilitate the studies on the ionospheric model and the determination of the
  • atmospheric water vapour content
  • Application for the geodynamic and scientific studies
  • Combination of the GPS derived ellipsoidal heights with a precise geoid model to replace conventional leveling.
3.1 Establishment of Dubai Virtual Reference Network
The Dubai Virtual GPS Reference System (DVRS) consists of five permanent GPS stations Figure3.1 and a Central Processing Centre for the processing and distribution of GPS data. This project is carried out by the Survey Section of the Dubai Municipality. The five permanent GPS stations within DVRS are continuously tracking all visible satellites. They were designed and constructed over the 2001 fiscal year. This network of permanent GPS stations will form the Zero-order geodetic network for the Dubai Emirate and connected to the ITRF epoch 2000. In all these permanent GPS stations, Leica receivers were installed. The GPS receivers will telemetry the data to the Central Processing Center via a GSM media at a measurement interval of 1 second. The core product of the DVRS is the data collected from the permanent GPS stations:
  • Daily 24-hour GPS carrier phase and code observations, on both frequencies, for all satellites in view
  • GPS navigation messages and status information.
The products that could be derived are:
  • Highly precise GPS satellite ephemeris
  • Earth rotation parameters
  • Ionospheric and atmospheric information
  • Coordinates and velocities of the permanent GPS stations.
3.2 Network Design
The listed conditions are minimum conditions consider in designing the DVRS Network: Network coverage of RTK services of whole Dubai Emirates.
  • Stable site (minimal local horizontal and vertical movement).
  • Stable antenna mount.
  • Minimum electromagnetic interference.
  • degrees located as far away from the reference antenna as possible and
  • located to the north of the reference antenna.
  • Adequate security for equipment.
  • Receiver and communications hub located inside a building
  • Providing protection from weather and elements.
  • Antenna located in a minimal Multipath environment.
  • Continuous long-term operation.
  • Availability of power supplies and telecommunication connection.

Figure 3.1 : DVRS Stations Distribution

3.3 Site Selection
The preliminary planning to select reference stations from the primary existing control stations but due to security and safety requirements we decided to find out locations where security and safety are available in additions to other requirements like securing the location of stations from any future replanning by coordination with planning and road departments. Three Base stations are situated on existing building (Al Quesis, Jabal Ali and Hatta) and the other two stations at Lusayili, Marqab developed and constructed from scratch

3.4 DVRS System Configuration

3.4.1 Hardware Configuration at Reference Station
The following hardware configuration at Reference Station are as follows :
  • GPS Antenna
      Leica AT 504 choke-ring (Figure 5.4), the system includes an International GPS Service (IGS), |Chock-Ring antenna designed by NASA/ Jet Propulsion Laboratory (JPL) with Multi-Path facility, it was calibrated by Geo++. Also covered by Radom cover for weather protection (Figure 3.4)
  • GPS Receiver
      Leica MC 500 (Figure 5.6), 12 channels L1 C/A or P-code. 12 channels L2 P-code, Full-wave length L1, L2 Carrier Phase.
  • Al CATELE MODEMS
      For sending raw data from Base-Station to Main-Control Room through Leased lines
3.4.2 Hardware Configuration at Main-Control Room
Hardware configuration being considered at MCR consist of :
  • 3 Personnal computers connected in network.
  • MOXA to convert serial port R232 into TCP-IP
  • 5 Modems for receiving raw data from 5 RS.
  • IPR.
  • Router to receive 30 calls simultaneously through IPR.
( IP-cluster software function is to make the raw GPS data coming from reference stations avialbel for two PC in Real Time through virtual IP adress)



3.4.3 DVRS Software
The software used in the DVRS is the GEO ++ Software known as Global Navigation Satellite System - State Monitoring And Representation Technique (GNSS-MART) Software (Wübbena, G., A. Bagge, Martin , Schmitz,2001).

GNSS-MART comprises the monitoring system including regional atmospheric effects, together with its presentation and delivery to the user for the purpose of position determination with highest accuracy, reliability and availability, both in real time and by post processing. GNSS-MART consist of the following list of software:
  • Reference stations (GNREF) Software
  • Communications systems (GNCOM, RTCM_IN, RTCM_OUT) software
  • Multi station solutions (GNNET) software
  • Real time applications (FKP area corrections, RTCM 2.1) software
The advantages of GNSS-MART could be listed as:
  • Capability of Networking with spacing more than 50km to enable position fixing
  • with centimeter. accuracy in Real Time and Post Processing.
  • Can be used for both GPS and GLONASS systems.
  • Error modeling of the system for satellite orbit, ionosphe troposphere.
  • Reduction of Multi-Path effects.
  • Elimination of Antenna Phase Center variations by antenna calibrations.
  • GNNET processes correction signals of several PDGPS reference stations in the RTCM 2.1 format such as they are created by GNREF.
  • Communication between the reference stations via modem connection, via transparent network connections (e.g. via Ethernet TCP/IP, ISDN-Routing) or via the normal RTCM-correction data signal (e.g. 2m radio).
  • Simultaneous processing of Five reference stations, thus, far-reaching registration of distance dependent errors in satellite orbits, ionosphere, troposphere.
  • Definition of correction models and parameters to describe these errors.
  • Generation of correction parameters for an extended RTCM,virtual reference stations(VRS),Pseudo Rereference Station (PRS) or Area Correction Parameters (FKP).
  • On the mobile station the additional correction parameters make the computation of optimal position dependent correction data possible.
  • GNNET increases the redundancy of the overall system by using several reference stations.
3.5 ITRF Connection
In 1995 during the DUREF-95 GPS campaign jointly conducted by Dubai Municipality and IfAG (BKG), GPS observation and computation have been carried out in order to connect the existing Dubai Geodetic Network to the International Terrestrial Reference System (ITRS) ITRF93. Four (4) stations from the DUREF-95 have been re-observed simultaneously with the DVRS stations on 11th May 2002 for six hours. With the connection, Helmert transformation parameters can be derived in order to established the relationship between ITRF93 and ITRF2000 coordinates.

The GPS data was processed using Bernese 4.2 software with the strategy similar to the DVRS permanent stations processing. Five (5) DVRS stations have been processed together with four (4) established triangulation stations (ITRF93) namely ET145, ET228, OBP5 and ET152.

Page 2 of 3
| Previous | Next |