We are IntechOpen, the first native scientific publisher of Open Access books. International authors and editors. Our authors are among the TOP 1%

Transcription

1 We are IntechOpen, the firt native cientific publiher of Open Acce book 3, , M Open acce book available International author and editor Download Our author are among the 151 Countrie delivered to TOP 1% mot cited cientit 12.2% Contributor from top 500 univeritie Selection of our book indexed in the Book Citation Index in Web of Science Core Collection (BKCI) Intereted in publihing with u? Contact book.department@intechopen.com Number diplayed above are baed on latet data collected. For more information viit

2 7 Precie Real-Time Poitioning Uing Network RTK Ahmed El-Mowafy Curtin Univerity Autralia 1. Introduction In the claic RTK method uing a ingle reference tation the rover need to work within a hort range from the reference tation due to the patial decorrelation of ditance-dependent error induced by the ionophere, tropophere and orbital error. The operating range of RTK poitioning i thu dependent on the exiting atmopheric condition and i uually limited to a ditance of up to km. In addition, no redundancy of the reference tation i uually available if the reference tation experience any malfunctioning. The contraint of the limited reference-to-rover range in RTK can be removed by uing a method known a Network RTK (NRTK), whereby a network of reference tation with range uually le than 100 km i ued. The network tation continuouly collect atellite obervation and end them to a central proceing facility, at which the tation obervation are proceed in a common network adjutment and obervation error and their correction are computed. The obervation correction obtained from the network are ent to the uer, operating within the coverage area of the network, to mitigate hi obervation error. In thi chapter, the principle of Network RTK are firt dicued and the advantage and diadvantage of the method are given. Next, the network deign parameter are dicued, which include network baeline length and configuration, the communication method between the computing centre and the uer, and the amount of calculation required by the network proceing centre and by the uer. Decription of poible network proceing technique, their baic model, and a comparion between their advantage and diadvantage are given. Finally, ome important NRTK application are dicued including the ue of NRTK in engineering urveying, machine automation and in the airborne mapping and navigation. Reult from real-time teting are dicued. 2. Principle of the network RTK The aim of network RTK i to minimie the influence of the ditance dependant error on the computed poition of a rover within the bound of the network. NRTK provide redundancy of reference tation in the olution, uch that if obervation from one reference tation are not available, a olution i till poible ince the obervation are gathered and proceed in a common network adjutment. Figure 1 illutrate a imple demontration of the concept of NRTK through repreentation of the relationhip between

3 162 Global Navigation Satellite Sytem Signal, Theory and Application the modelled ditance-dependent error and their actual value. The error plane at the three hown reference tation are at different level. The NRTK provide an error urface formed from the error at the three reference tation (a plane in thi cae). The actual change of error between the reference tation i hown in red. If a uer i cloe to any of the tation, auming having the ame level of error of that reference tation will give reaonable accuracy and reult in mall poitioning error at the rover. A the uer move away from the reference tation, the magnitude of the differential error between the actual and the reference tation error level increae. On the other hand, the differential error between the actual error and the NRTK etimated error, interpolated on the NRTK error urface at the location of the rover when ued, i ignificantly minimied. In principle, the RTK network approach conit of four baic egment: data collection at the reference tation; manipulation of the data and generation of correction at the network proceing centre; broadcating the correction, and finally poitioning at the rover utilizing information from the NRTK. In the firt egment, multiple reference tation imultaneouly collect GNSS atellite obervation and end them to the control centre, where a main computer directly control all the reference tation, motly via the Internet. All reference tation hould ue geodetic-grade multi-frequency GNSS receiver. The incoming GNSS obervation data from all operating reference tation are creened for blunder and next their ambiguitie are fixed. The control computer ue thee data in proceing a networking olution, and the data are archived for pot-proceing ue. The network information are then broadcat to uer. The network information depend on the proceing algorithm and may include any of the following: obervation from one reference tation (phyical or virtual), coefficient for interpolation of correction within the coverage area, and obervation correction at a group of reference tation. To increae reliability, it i recommended to let a econd computer work in real time a a backup to the main computer in the event of any malfunctioning. Ref1 NTRK error urface (plane for 3 tation) Error error at Ref1 change of actual error actual error at rover Rover Ref2 Ditance error at Ref2 Fig. 1. Relationhip between error in a mall NRTK coverage area Ref3 error at Ref3

4 Precie Real-Time Poitioning Uing Network RTK 163 NRTK uually require a minimum of three reference tation to generate correction for the network area. In general there i no retriction concerning the network ize, it can be regional, national, or even international. However, reference tation eparation i uually retricted to le than 100 km to allow for quick and reliable ambiguity reolution. A the number of tation increae, redundancy increae, and better correction can be etimated. If one or two reference tation fail at the ame time, their contribution can be eliminated from the olution and the remaining reference tation can till provide the uer with correction and give reliable reult (El-Mowafy et al., 2003, Hu et al., 2003). Typically, a NRTK erver ytem would conit of the following component (e.g. Leica Geo ytem, 2011): A ite erver connected to each reference tation receiver, A network erver that acquire the data from the ite erver and end it to the proceing centre, A cluter erver that hot the network proceing oftware. The oftware perform everal tak including: quality check of data, apply antenna phae centre correction, ambiguity fixing, modelling and etimation of ytematic error, interpolation of error (correction) in ome technique (e.g. VRS, PRS) and generation of virtual obervation, model coefficient in other (FKP), or Mac data. A firewall i uually etablihed to protect the above erver from being acceed by a uer. RTK proxy erver to deal with requet from the uer and end back network information. The uer interface to end/receive data from the NRTK centre. The main advantage of the Network RTK can be ummaried a follow: Cot and labour reduction, a there i no need to et up a bae reference tation for each uer. Accuracy of the computed rover poition are more homogeneou and conitent a error mitigation refer to one proceing oftware, which ue the ame functional and tochatic modelling and aumption, and ue the ame datum. Accuracy i maintained over larger ditance between the reference tation and the rover. The ame area can be covered with fewer reference tation compared to the number of permanent reference tation required uing ingle reference RTK. The eparation ditance between network tation are ten of kilometre, uually kept le than 100 km. NRTK provide higher reliability and availability of RTK correction with improved redundancy, uch that if one tation uffer from malfunctioning a olution can till be obtained from the ret of the reference tation. Network RTK i capable of upporting multiple uer and application. Network RTK ha though ome diadvantage, which are: The cot of ubcription with a NRTK provider. The cot of wirele communication with the network (typically via a wirele mobile uing for intance GPRS technology). The dependence on an external ource to provide eential information.

5 164 Global Navigation Satellite Sytem Signal, Theory and Application 3. Network deign parameter Etablihing a network RTK uually tart after a thorough cot/benefit analyi. At the deign tage, the following main factor hould be conidered: 1. Baeline length (ditance between the reference tation) tation location and network configuration (number and geometric ditribution of reference tation). 2. The communication method between the computing centre and the uer. 3. Calculation required by the network control and by the uer (network algorithm). Thee factor are dicued in the following ection. 3.1 Ditance between the reference tation and network configuration The main advantage of the network approach i that it improve modelling of the ditance-dependent error over long ditance (El-Mowafy et al., 2003 and Euler et al., 2004). The obervation correction (computed a the ame value of the error but with oppoite ign) can be generated after removing cycle lip and determining double differenced phae ambiguitie between the reference tation. A major technical challenge in NRTK i ambiguity reolution within a reaonable hort period over uch large ditance between reference tation. In order to achieve a fat and reliable ambiguity reolution, the ditance between the reference tation i better choen not to exceed 100 km (Wübbena and Willgali, 2001). Typically, baeline length in NRTK range between 20 km and 100 km (70 km on average). In principle, a minimum of three tation i required to generate RTK network correction, but in practice thi number hould not be le than five. The increaed redundancy of reference tation improve poitioning accuracy and ambiguity reolution and help to utain network availability and reliability in the cae of a temporary failure of any reference tation. However, the degree of redundancy hould be evaluated by mean of a cot/benefit analyi, balancing the need to improve the economical apect of etablihing and running the network and keeping the required degree of redundancy. Hence, election of baeline length by the network deigner hould atify the following condition: covering of the whole area of interet with reliable correction; maintaining ufficient tation redundancy; achieving a reliable ambiguity reolution with acceptable confidence level at every location within the network area a long a a minimum of five atellite are oberved; Enuring reliable communication between the reference tation and the network centre (motly via the internet through land line, and in remote area through atellite communication); chooing ite free from multipath and radio frequency interference. It i alo preferable to have the reference tation ituated at a imilar altitude. In network configuration, the following can be taken into conideration: For a limited number of reference tation, it i recommended to hape the network a a polygon with one or more central tation. A compact hape of the network i preferable (i.e. a circular network i better than a rectangular network).

6 Precie Real-Time Poitioning Uing Network RTK 165 Some geographic region (i.e. equatorial or high latitude) will require a dener network than in the mid latitude due to poorer-atellite geometry, atellite availability, and ionophere diturbance etc. In practice, a main factor affecting the choice of tation ditance and network configuration i finding uitable ite for the reference tation. The main conideration are availability of communication infratructure and obtaining approval of ite owner (whether a government or a private ector). It i alo poible to integrate obervation correction etimated from network of different ize. For intance, error of regional or even global nature, uch a atellite orbital error, clock and the regional behavior of the ionophere, which lowly change, can be etimated from regional network. The local ionopheric and tropopheric error on the other hand can be etimated from local network. Thu, RTK network can be configured uch that area of heavy uage can be covered by a cloe-mehed reference tation network for highet accuracy and reliability in poitioning, wherea le important area are covered by a widemehed network of regional or national extenion (Wübbena et al., 2001). 3.2 Communication method between the proceing centre and the uer Real-time application require a communication link between a ervice provider and the uer. Currently, there are two main mode of communication that can be ued in network RTK; either a duplex (bi-directional) communication or one-direction communication. Each method ha it advantage and diadvantage. In chooing which communication method to ue, the deigner ha to conider economical factor uch a: operational cot by the uer, cot of maintenance of exiting infratructure and/or building new one, and the amount of computation needed by the rover and the proceing centre. The technical apect that need to be addreed include: expected ignal trength at different location, number of uer, range and coverage (Wu, 2009), tranmiion bandwidth, protocol, reliability and error correction, latency (one econd and horter data tranmiion latencie are required for cm level poitioning accuracy). At preent, the duplex communication mode i the motly ued method. In thi mode, a cellular modem uch a a General Packet Radio Service (GPRS) or Global Sytem for Mobile Communication (GSM) are ued. GPRS i uually preferred a it i more economical than GSM ince the uer only pay for the data packet received, not for the entire call duration when uing GSM. GPRS can provide a table and reliable connection with latencie le than one econd (Hu et al., 2002). The duplex approach ha a retriction on the number of uer, a thi number i limited by the ability of the NRTK proceing centre to imultaneouly perform calculation for all uer. Thi may alo reult in extended latency in receiving the network information. For a limited number of uer thi latency i uually le than three econd. On the other hand, the one-direction communication method mainly employ VHF or UHF broadcating or encode the RTK correction into a broadcat TV audio ub-carrier

7 166 Global Navigation Satellite Sytem Signal, Theory and Application ignal (Petrovki et al., 2001). For VHF broadcating, allocation of uitable broadcat radio frequency and obtaining it licene i an important iue in the early development of a network RTK. The main advantage of thi method i that there i no retriction on the number of uer concurrently uing the NRTK ervice. However, the main diadvantage of the method i the high cot of the infratructure needed to build radio ignal repeater, if needed, to cover the whole area. In addition, ome problem can be experienced due to the poibility of receiving ignal of varying trength in different location, and poible frequency jamming. A mix of both communication method i however poible (Cruddace et al., (2002). The data tranmiion from the reference tation to the control centre erver and from the control centre erver to the uer for RTK correction i motly carried out via the Network Tranport of RTCM via Internet Protocal (Ntrip), BKG, Ntrip i an open ource and can be downloaded from the internet (LENZ, 2004). Ntrip wa built over the TCP/IP foundation and i an application level protocol for treaming GNSS data over the internet. It wa firt developed by the German Federal Agency for Cartography and Geodey (BKG). Ntrip ue HTTP and ha three component: Ntrip Client, Server and Ntrip Cater. Ntrip i deigned for dieminating differential correction data (e.g. in the RTCM-SC104 format) or other kind of GNSS treaming data to tationary or mobile uer over the internet. It allow imultaneou PC, Laptop, PDA, or receiver connection to a broadcating hot. Ntrip upport wirele internet acce through mobile IP network like GSM, GPRS, EDGE, or UMTS (BKG, 2011). To reduce latency, the amount of data tranmitted to the rover hould be minimied. One poible olution i to change (optimie) the update rate for the different parameter to follow their phyical behaviour. Ditance dependent error can thu be eparated into a diperive component, coniting mainly of the ionopheric refraction, and a non-diperive component coniting of the tropopheric refraction and orbit error. Different propoal for optimiing the update rate have been made. An update rate of 15 econd eem reaonable for non-diperive correction difference, while an update rate of only 10 econd may be ufficient for the diperive contribution (Euler et al., 2004). However, the impact of thee rate on the Time-To-Firt-Fix (TTF) of carrier phae ambiguitie hould be carefully tudied, a it lie at the top of the uer interet (El-Mowafy, 2005). The type of communication ued alo affect the network algorithm and the amount of calculation required at the proceing centre and by the uer. For intance, if a bidirectional communication i ued, the proceing centre can individualie the network information for a uer baed on hi/her approximate location. Thu, the computation made at the uer receiver are minimied. On the other hand, if the data link i one-directional, the uer ha to make the neceary interpolation of error at hi location and ha to identify a uitable reference tation to ue. 3.3 NRTK olution method Currently everal olution method can be applied in Network RTK, including the Virtual Reference Station (VRS), Peudo-Reference Station (PRS), individualied Mater-Auxiliary correction (imax), Area-Parameter Correction (Flächenkorrekturparameter FKP- in it German origin), and the Mater-Auxiliary (MAC) method. In VRS, PRS and imax

8 Precie Real-Time Poitioning Uing Network RTK 167 referencing i made to a non-phyical reference tation located in the vicinity of the approximate poition of the rover and virtual obervation are generated to refer to thi non-phyical reference tation. The uer typically ha no information about the ize of error and their behaviour. In contrat to the non-phyical network approach, FKP and MAC broadcat raw reference tation obervation and network information eparately. The network information i repreented by diperive and non-diperive correction and the rover oftware decide how the network information i applied. A ummary of thee method i given in ection 5. Once the network error are computed at the reference tation, ditance-dependent error need to be interpolated at the location of the uer receiver. Several method can be ued for uch interpolation proce including: the ue of linear interpolation, uing a linear combination model, applying an invere-ditance linear interpolation or a low-order urface model (ued for example in the FKP technique), utiliation of the leat-quare collocation approach, or uing Kriging technique (ee for intance Fototpoulo, 2000, Dai et al., 2001, Wu, 2009 and Al-Shaery et al., 2010). 4. Etimation of the diperive and non-diperive error at the network reference tation The mathematical equation of the code and phae obervation for the receiver (j) and the atellite () at time (t) can be written a: Where: j j j R j j = j + δ + δ j δ + j ( ) +δ j ( ) j ( ) δ j ( ) + j +εpj Rj C R r c( t t ) T t T t I t I t p (1) R j c j Rj r c (t j t) T(t) j T(t) j I(t) j I(t) j Nj pi j R f j φ = + δ + δ δ + +δ δ + + +ε φ (2) C, φ code and phae obervation, repectively; R geometric range between the uer antenna and the atellite; δr orbit error; c peed of light; δt j, δt receiver and atellite clock error, repectively; T j (t) modelled tropopheric refraction delay (mainly the hydrotatic, dry component of the tropophere); δ Tj () t reidual tropopheric refraction delay (mainly the unmodelled wet tropophere); Ij ( t ) modelled ionopheric refraction delay if applied (frequency dependent); δ Ij ( t) reidual ionopheric refraction (frequency dependent); f ignal frequency; N j integer phae ambiguity; p j total ite dependent error (antenna phae centre and multipath δ M j ); ε, ε code and phae remaining random noie, repectively. P j φj

9 168 Global Navigation Satellite Sytem Signal, Theory and Application From the above equation, one can ee that poitioning accuracy from GNSS phae obervation i limited by two type of error: the ditance dependent error, which include orbit, ionophere and tropophere error, and tation dependent error, which include multipath, antenna phae centre variation, and receiver hardware biae. The network etimation methodology ue the known information of the antennae and ite to reduce tation related error and focue on etimating the ditance-dependent error. For the tation-dependent error, multipath can be minimied uing choke ring and modelling the ite pecific multipath pattern taking advantage of the fixed reflector to antenna geometry at reference tation and of the daily repeatability of multipath. Thi can be done utilizing technique uch a the Hilbert Huang tranformation to decompoe the time-hifted pot-fit GPS phae ignal reidual (Hieh and Wu, 2008). Another approach i to include multipath error in the network etimation proce, which will average out the uncorrelated multipath error. To minimie the antenna phae centre variation, the definition of the network reference tation antennae ha alway to be conitent. Thi can be done by uing the ame antenna model type for all reference tation and unifying antenna orientation. To eliminate the phae centre variation, an abolute calibration of each antenna i recommended. However, mot current network only apply relative calibration of the antennae, which i a tandard calibration proce that can be applied for the type of antenna ued, determined relative to a reference antenna (typically a Dome Margolin Model T with choke ring). The ditance dependent error can be eparated into a diperive component (i.e. frequency dependent), which i the error induced by the ionophere, and a non-diperive component, that include orbital and tropopheric error. Etimation of the diperive and non-diperive error at the network reference tation can be performed in everal way. In one approach the tate of individual GPS error in real time can be etimated by proceing all tation of the network imultaneouly uing un-differenced obervable (Wübbena and Willgali, 2001, Zebhauer et al., 2002, Wübbena et al., 2005). Then, the tate vector ( X ) at tation j read: X=N, t, t, r, T, r ( ) T j δ j δ δ δ j δ j I, δmj (3) The orbital and tropopheric error are combined to form the geometric (non-diperive) error δr, the ionopheric diperive error δr (replacing the term I I j and δi j and in Equation 1 and 2). The tate pace approach ha ome advantage; the main one i it ability to contrain each bia by pecific model (Wübbena and Willgali, 2001). Alo, a change in the network configuration caued by the breakdown of one of the reference tation can be compenated without much effort. Moreover, in the cae of irregular condition of one of the tate parameter, warning can be iued to the uer. Another popular method for etimation of network error i uing ingle difference linear combination of obervation. The diperive and non-diperive component are determined for atellite and between the reference tation j and k, uing dual-frequency receiver of L1 and L2, a follow: δ r I, δ r L δ r L (4)

10 Precie Real-Time Poitioning Uing Network RTK 169 δ r, δ r L δ r L (5) 5. Summary of network RTK proceing technique In thi ection, the mot common network RTK technique ued at preent are dicued, namely: the VRS, FKP, and Mac. 5.1 The Virtual Reference Station (VRS) method The VRS technique i currently the mot popular NRTK method due to the fact that it doe not require change in the uer oftware, i.e. it i compatible with exiting oftware. The rover applie the tandard differential poitioning of it obervation with obervation from a virtual reference tation. The ditance-dependent error are computed for each pair of atellite, and for each mater-to-another-reference tation. The VRS method require bidirectional communication. The rover end it approximate poition via a wirele communication link (typically a cellular modem in NMEA format) to the network proceing centre where computation are carried out for each uer (Vollath et al., 2000, Hu et al., 2003). Some network provider ue only the nearet three-to-five reference tation to compute the meaurement error for a pecific uer, ee Figure 2. The etimated network meaurement ditance-dependent error are interpolated for a virtual reference tation (VRS). The VRS location i typically elected at the initial approximate poition of the rover. For a kinematic uer, thi VRS location i kept to preerve the ambiguity value determined from it olution until the range between the VRS and the actual poition of the rover become too long for precie differential poitioning. Then, a new VRS at the mot recent poition of the uer i etablihed. ref t. ref t. j VRS i rover ref t. NRTK centre Fig. 2. VRS concept To contruct the obervation at the VRS, the VRS and the atellite known poition are firtly ued to compute the range between the atellite and the VRS. Similarly, the range between the atellite and the mater tation i computed, where the mater tation i uually

11 170 Global Navigation Satellite Sytem Signal, Theory and Application elected a the nearet continuouly operating reference tation to the uer. The range difference between the VRS (point i) and mater tation (point j) with repect to atellite () read: R R R (6) Where: R R R atellite to VRS range atellite to the mater tation range range difference The interpolated ditance-dependent error, diperive δ r and non-diperive δ r I or their total, at the VRS are added to the mater tation obervation to generate the VRS obervation on a atellite-by-atellite and epoch-by-epoch bai for L1 and L2 frequencie, uch that: R δ r δ r I T /λ (7) P P R δ r δ r I T (8) where ΔT ji repreent the difference in the modelled part of the tropophere, which i uually ubtracted before computation of the network error, and thu need to be reconidered. Previou teting of the VRS ytem for kinematic ground urveying howed that ytem poitioning accuracy wa typically 1-2 cm in plane coordinate and 3-5 cm in height (El-Mowafy et al., 2003). The raw meaurement or their correction are ent to the rover in RTCM verion 2.1 format uing meage type 18/19 or 20/21, repectively. If the former type are ued, the contructed VRS obervation are computed by the network centre wherea if the latter meage type are ued, the uer ha to compute the VRS obervation. A variation to the VRS concept i the Peudo Reference Station approach (PRS), where the virtual reference tation i taken at a pre-elected grid point intead of the approximate poition of the uer. The virtual obervation in thi cae will alo refer to a non-phyical reference tation. At tart of the urvey, the baeline length i typically a few metre for the VRS approach and could be everal kilometre for the PRS method. The baic advantage of the VRS mode are that it doe not need oftware change in the uer receiver, and no pecial format and convention are needed. However, a main drawback of the method i the preence of a retriction on the number of uer according to the capacity of the network proceing centre due to the fact that VRS obervation are cutomied for each uer. For kinematic application, re-determination of VRS may be needed according to the ditance between the uer and the VRS. In addition, if RTCM meage type 18/19 are ued, the uer will have no information about error ize, which alway help in interpretation and analyi of poitioning reult. 5.2 The Area-Parameter Correction-Flächenkorrekturparameter (FKP) method The Flächenkorrekturparameter (FKP) or Area-Parameter Correction method repreent the network information uing coefficient of a urface centred at the location of a phyical reference tation (Wübbena and Bagge, 2006). Raw reference tation obervation and

12 Precie Real-Time Poitioning Uing Network RTK 171 network information are broadcat eparately. The method require advanced oftware in the rover receiver to do interpolation of correction. However, unlike the VRS approach, the uer ha information about error ize, which help in quality control and analyi of reult. The rover oftware decide how the network information i applied. For intance, the uer can apply the correction at hi location to mitigate obervation error and do differential poitioning with the broadcat mater tation data. Alternatively, the uer can utilie the network information to contruct a VRS at a nearby location, or the uer may apply the Precie Point Poitioning (PPP) in what i known a PPP-RTK (Teunien et al., 2010). FKP can apply an open meage with one-directional communication (from centre to uer) to cover a certain area. In thi cae, no retriction would exit on the number of uer. A bidirectional communication can alo be employed. In FKP, the reidual at the network reference tation are aumed to define a urface which i parallel to the WGS-84 ellipoid in the height of the reference tation. For baeline le than 100 km, the patial variation of the reidual can be approximated by a low-order urface model, e.g. a plane, uing a bilinear polynomial in the form: where: δr(t) = a(t) (φ φ R ) + b(t) (λ - λ R ) + c(t) (9) a, b, c coefficient defining the plane at time (t). a and b model the trend of change of error within the area, and c i ued for modelling the tation pecific error of the mater tation if undifferenced obervation are ued or the averaged value of the tation pecific error of all the tation if double difference obervation are ued (Wu et al., 2009). φ, λ geographic coordinate of the interpolated point (in radian) φ R, λ R geographic coordinate of the reference point (in radian) The coefficient are etimated from a weighted leat quare olution from the computed reidual at each reference tation uing Equation 3 or 4 and 5. For intance, for n number of reference tation we have: 1 1 (10) 1 where Δλ R-j and Δφ R-j are the difference in latitude and longitude between the reference tation R and tation j, repectively. The leat-quare etimate for the coefficient can be obtained by (Wu, 2009): Where: (11) 1 1 and (12) 1

13 172 Global Navigation Satellite Sytem Signal, Theory and Application Typically in FKP, two plane are computed for each atellite, centred at each reference tation, one plane for the diperive and another for the non-diperive correction. The correction at the rover are determined through interpolation uing the inclination parameter of the correction plane. Reult of Euler et al., 2002, howed that the plane urface model gave good reult when modelling the regional trend of the correction difference. Although low-order (linear-plane) urface model are uually utilied, longer baeline between reference tation may require polynomial of a higher order. The urface area model wa dicued in everal publication, e.g. Varner, 2000, Fotopoulo and Cannon, 2001 and Wübbena and Bagge, An example i given by the latter tudy for generation of FKP, where the error are computed a follow: Where: δr δr I FKP NI FKP EI FKP N FKP E θ δr o (t) = 6.37 (FKP N (φ φ R ) + FKP E (λ λ R ) co(φ R )) (t) (13) δr I (t) = 6.37 α (FKP NI (φ φ R ) + FKP EI (λ λ R ) co(φ R )) (t) (14) α = (0.53 θ/π) 3 (15) etimated non-diperive geometric (orbital and tropopheric) error etimated diperive ionopheric error The FKP parameter in north-outh direction for the ionopheric ignal narrow lane in ppm The FKP parameter in eat-wet direction for the ionopheric ignal narrow lane in ppm The FKP parameter in north-outh direction for the geometric ignal ionopherefree in ppm The FKP parameter in eat-wet direction for the geometric ignal ionopherefree in ppm the atellite elevation angle in radian After interpolating the diperive and non-diperive error, they are combined to generate the range reidual for L1 and L2 frequency obervation, which read: Where: δr δr δr I (16) δr δr δr I (17) δr, δr total meaurement error for the frequencie f 1 and f 2. f 2, f 1 frequencie of L 1 and L 2 ignal. Finally, the range R, derived from the carrier-phae meaurement are corrected a follow: R R δ r R R δ r (18) (19)

14 Precie Real-Time Poitioning Uing Network RTK 173 The drawback of the FKP method include the need of the rover to perform interpolation of meaurement correction, poible inconitency at the edge of two adjacent plane due to the ue of the linear plane urface, and large data format are needed. In Radio Technical Commiion for Maritime ervice (RTCM) format verion 3.1, FKP correction can be ent via meage type 1034 and 1035 for GPS and GLONASS obervation, repectively. 5.3 The Mater-Auxiliary (MAC) method In the Mac approach, the rover end it approximate poition via NMEA format to the network proceing centre. The centre determine for thi pecific uer the appropriate mater tation, which i uuall elected the cloet reference tation to it poition, and identifie the auxiliary reference tation. Thee tation are choen within a catch circle of a predefined radiu (e.g. 70 km) around the rover, and with a pre-et number (e.g. from 3 to 14). Figure 3 illutrate the Mac concept. In one Mac approach, a network RTK of large number of reference tation can be ubdivided into cluter (Leica Geo ytem, 2011). The proceing centre define the appropriate cluter to a uer and define the appropriate network correction applicable to that uer. Mater j Aux k 4 Auxiliary k 1 Fig. 3. The Mater Auxiliary NRTK The rover can receive different type of information according to the trategy ued by the Mac proceing centre which may include: The coordinate and raw meaurement of the Mater tation. Meaurement correction at the Mater tation. Correction difference between the Mater and Auxiliary tation. Thee difference when being added to the correction of the Mater will give the correction at the Auxiliary tation. The latter Mac correction can be received via RTCM v3.1 meage type , and etc. In Equation (7 and 8), the ingle obervation difference between the Mater tation j and an Auxiliary tation k for atellite read (Takac and Lienhart, 2008): R rover Aux k 2 Aux k 3 δ r δ r I T /λ (20)

15 174 Global Navigation Satellite Sytem Signal, Theory and Application P P R δ r δ r I T (21) One characteritic of the Mac approach i that it data are ent to the uer at the ame ambiguity level. Thi can be explained a follow. In the Mac method the carrier phae ambiguitie are determined with repect to fixed ingle difference ambiguity value. However, ambiguity fixing i more reliably performed uing a double difference approach. Thu, the ambiguitie from the atellite can be determined from that of the reference atellite q and their double difference a follow: N N, N (22) Therefore, the ambiguity bia, which i the difference between the true ambiguity and the etimated ambiguity for the reference atellite, uually known a the ambiguity level, hould be etimated. It i common to all etimated ambiguitie of atellite oberved from one baeline and cancel out in double differencing. After receiving the Mac information, the rover oftware i free to decide the method of interpolating the correction at it location. The proceing centre can do the interpolation if needed (individualied I-Max). The rover oftware i alo free on how the Mac information be ued to determine it poition. For intance, the rover can apply double differencing with the Mater reference tation a the bae. It can alo do that after removing the error from both the Mater reference tation and it poition. 6. PPP- RTK A more recent direction of NRTK implementation i it integration with the precie point poitioning (PPP) technique, Wübbena et al., In a tandalone PPP mode, undifferenced obervation are ued and the atellite related error are mitigated by uing atellite clock correction and utiliing precie orbit to avoid the orbital error aociated with the ue of broadcat ephemeri. Thee atellite product are typically provided from a proceing centre analying global data uch a the International GNSS Service (IGS). Since only one receiver i ued in PPP, the ambiguitie are olved a part of the unknown with real number and not fixed. A a reult, everal minute of data are needed when proceing to achieve a reliable convergence of the olution. A the ambiguitie are olved a real number, only accuracy at the ub-decimetre, at bet, i achievable from PPP. However, it i poible to integrate PPP and NRTK into a eamle poitioning ervice, which can provide an accuracy of a few centimetre (Li et al., 2011). The concept of PPP-RTK i to augment PPP etimation with precie un-differenced atmopheric correction and atellite clock correction from a reference network, o that intantaneou ambiguity fixing i achievable for uer within the network coverage. A few technique have been yet propoed for PPP-RTK. In the method preented in Teunien et al., 2010, un-differenced obervation equation for the network tation are ued, and thu the deign matrix of the network will how a rank defect. Thi rank defect i eliminated through an appropriate reparametrization (i.e. reduction and redefinition of the unknown parameter). Thi reult in redefined atellite clock and ambiguitie. The tropopheric delay i lumped with the phae and peudo range atellite clock error and the ambiguity become a between receiver ingle-differenced ambiguity. Eventually, a full-rank ytem of obervation can be obtained.

16 Precie Real-Time Poitioning Uing Network RTK 175 In PPP-RTK, the function of the network i to provide the uer with atellite clock and interpolated ionopheric delay. When thee precie etimate are paed on to the uer, the above given definition of thee clock enure that the ambiguitie of the uer are alo integer and ambiguity reolution i available at the uer ide. Satellite clock for each epoch are added a peudo-obervation, with appropriate variance matrix. The precie IGS orbit are ued. For the network proceing in Teunien et al., 2010, a Kalman filter i ued, auming the ambiguitie are time-invariant, while for the uer, an epoch-by-epoch leatquare proceing i ued, thu providing truly intantaneou ingle-epoch olution. The integer ambiguity reolution of both network and uer i baed on the LAMBDA method (Teunien, 1995), with the Fixed Failure Ratio Tet (Teunien and Verhagen, 2009). 7. Network RTK application In thi ection important application that can benefit from the few cm-level poitioning preciion and accuracy achievable by uing a ingle GNSS receiver with NRTK are preented. 7.1 NRTK in engineering urveying Surveying work in contruction ite are uually dependent on determination of accurate coordinate and height. The 3D poitioning veratility and accuracy achievable from NRTK encourage the ue of thi technique for contruction urveying work, particularly for large ite when a rapid urvey i needed. The method help in reducing field expene and time due to reduction of the ize of urveying crew, elimination of the need for frequent etup of the urveying intrument, and the reduction of the need for accurate local travere or multiple control tation within the ite. Studie howed that RTK GPS and the traditional technique employing total tation gave tatitically compatible reult (El-Mowafy, 2000). With a typical accuracy of 1-5 cm, the NRTK GPS technique can be utilied for medium accuracy contruction urvey work uch a: - grading, - taking out of mark with medium accuracy, uch a road, footing, pipeline, utilitie, landcaping, fence etc., - cadatral urvey, - mapping, - checking of the a-built tructure, - ite exploration for new project. A NRTK GNSS ytem can be integrated with the total tation for intantaneou determination of the total tation location by mounting the GNSS antenna directly on top of the total tation alidade in open ite. Thu, the need for etablihing permanent horizontal control tation onite can be minimied. For orientation determination, the total tation can be ighted at a back tation, where it coordinate can be intantaneouly determined uing the NRTK-GNSS technique. Thi proce improve the economic of urveying work, and reduce the overall urveying time, including the time required for the initialiation of the total tation at each etup. However, one hould note that performance of urveying with RTK GNSS in contruction ite are affected by atellite availability, multipath error reulting from working near building, and latency of the reference data. The influence of

17 176 Global Navigation Satellite Sytem Signal, Theory and Application the number of atellite in view, dilution of preciion (DOP) and age of correction over the accuracy and tability of the NRTK GPS olution wa dicued in ome tudie, e.g. Aponte et al., The performance of urveying with the network RTK approach in contruction ite wa evaluated by two tet. The firt tet wa executed for checking performance in determination of planimetric coordinate and the econd tet for evaluation of performance in height determination. The firt tet wa carried out during contruction of a large building in Dubai for checking of poition of urveying mark of the footing, landcaping and the acce road of the building. 48 point were ued for checking purpoe including 18 point on the boundary of the road, 19 point for the footing, and 11 point for the landcaping. Thee point were et out uing a calibrated total tation of 1 econd preciion. Next, point coordinate were computed from the working drawing and uploaded to a GPS controller, where a ingle GPS dual-frequency receiver wa independently ued for poitioning of the tet mark by utilizing data from the Dubai NRTK, known a Dubai Virtual Reference Sytem (DVRS). The network conit of five continuouly operating reference tation with baeline length ranging from 23.4 km to 90.8 km and ue the VRS algorithm. The poition of each tet point wa determined after 10 econd of data collection, which were recorded at 2 econd interval. The hift between the poition determined from the two method, namely: calibrated total tation and NRTK uing obervation from GPS, were meaured and compared to repreent the preciion of the latter method if ued in contruction ite intead of the former method. Figure 4 illutrate the difference between the two method. The tatitic of coordinate difference between the two method in eating and northing are given in Table 1. The average norm of the patial difference ( ) wa generally le than 1.45 cm wherea the maximum difference wa 3.45 cm. The mall error can be explained by the preence of one of the network reference tation within a few kilometre, which would be picked up by the ytem a the mater reference tation. Thu, mot orbital and atmopheric error were cancelled and the remaining error would mainly be due to data noie and multipath. Thee reult how that the GPS-RTK network approach can be ued in the etting out of medium-accuracy urveying mark. Error in Northing Coordinate (cm) Error in Eating Coordinate (cm) Fig. 4. Poitioning difference between GPS Network RTK and the total tation

18 Precie Real-Time Poitioning Uing Network RTK 177 Average (cm) Maximum (cm) RMS (cm) E N Table 1. Statitic of poitioning difference between GPS RTK-Network with the total tation To check repeatability of reult (internal accuracy), another independent urvey wa carried out after two day and 4 hour to re-determine the coordinate of the ame marked point previouly determined by GPS. The average and maximum value of the difference between the reult of the two urvey are given in Table 2. A the table demontrate, repeatability teting howed that the average value for difference in the total planimeteric coordinate etimation wa at the cm level, while for ellipoidal height determination it wa 1.56 cm. Thee difference can be attributed to change in the quality of the meaurement ued, which mainly reulted from difference in the number of the oberved atellite and their geometric ditribution. Thee parameter have affected the quality of the network computation of the meaurement correction and the quality of coordinate etimation at the rover. Average Maximum (cm) (cm) E N h Table 2. Statitic of coordinate dicrepancie between different oberving eion Unlike traditional levelling, GPS derived height are referenced to an ellipoidal datum (WGS 84) and do not depend on local gravity variation, wherea in mot levelling work and mapping orthometric height are ued. Orthometric height reflect change in topography a well a local variation in gravity. They are referenced to the geoid, which i an equi-potential level urface of the Earth that i cloely aociated with the mean ea level on a global bai. To convert ellipoidal height from GPS (h GPS ) into orthometric height (H), geoid height are needed, uch that: H = h GPS - N (23) where (N) i the geoid height. Thu, with the ue of one receiver and employing NRTK to determine ellipoidal height, orthometric height can be determined if a good geoid model i available. To ae accuracy of orthometric height determination by uing NRTK, the econd tet wa performed in Dubai, uing the DVRS network. The Dubai gravimetric geoid model wa ued, which wa developed by integrating a comprehenive et of gravity meaurement with GPS, levelling and digital elevation data. The computed geoid fit GPS/levelling at the 3-4 cm level RMS (Forberg et al. 2001). The tet wa performed on a network coniting of 41 benchmark of the econd order levelling network. Orthometric height at thee

19 178 Global Navigation Satellite Sytem Signal, Theory and Application benchmark were firt etimated by combining ellipoidal height determined by uing the Dubai NRTK with the local gravimetric geoid model data and were next compared to known orthometric height of the benchmark. The tet area panned approximately 22.7 km x 7.8 km in the Eating and Northing direction repectively, repreenting the area acquiring the mot demanding urvey work in the Emirate of Dubai. The height difference between the highet and lowet point in thi tet wa approximately 34.5 meter. Each tet point wa occupied for a period of a few econd, repreenting an ordinary working environment. The tandard deviation of the ellipoidal height determination for the occupied point of the tet network ranged between 1.05 cm and 5.47 cm (El-Mowafy et al., 2005). Figure 5 how the difference in orthometric height between uing the NRTK GPS + geoid height and the known orthometric height of the benchmark. On average, difference were within ±5 cm, with a maximum value of 7.04 cm. The tatitical reult of the difference are preented in Table 3. The average value of the abolute difference wa 2.4 cm with 3.05 cm tandard deviation. The difference toward the north-eat were greater than thoe at the outh-wet region of the tet. Thi can mainly be attributed to accuracy of the geoid model ued in the tet area. Figure 6 illutrate the urface plot of the height difference between the two method. Thee reult how that no ignificant ytematic error were preent. The achieved accuracy i conidered precie enough for third order levelling, which repreent the majority of levelling work being carried out. H ([DVRS+Geoid] - Leveling) Point Fig. 5. Height difference between GPS Network RTK geoid height and precie levelling (cm) Average of abolute value Max. difference Table 3. Statitic of height difference between the GPS network RTK + geoid and precie levelling (cm) σ

20 Precie Real-Time Poitioning Uing Network RTK 179 Diff. (cm) Eating Northing Fig. 6. Surface plot of the height difference 7.2 Uing RTK GPS for remotely monitoring and controlling machine automation The ue of Real-time GNSS poitioning for machine automation can enhance it productivity and functionality. Real-time GNSS poitioning can provide cm accuracy, facilitating high performance and output for machine that require poitioning data. The Superviory Control and Data Acquiition (SCADA) i a good example for upporting a field automated ytem. Having real-time accurate GNSS poitioning information a an input to the SCADA ytem give a lot of opportunitie to develop many olution for planning, deign, contruction and monitoring field operation that require precie poitioning. An automated machine uch a a field tractor, excavator, or a driller can be automatically operated, unmanned and fully remotely monitored and controlled. The machine can be controlled by the analogue and digital output from a Remote Terminal Unit (RTU). The RTU can have preloaded Programmable Logic Control (PLC) oftware that activate the output according to the field input from the machine primary enor a well a the realtime accurate coordinate fed from a GNSS unit receiving meaurement correction from a NRTK centre. The field operation, event, log and alarm can be fully remotely monitored through a SCADA ytem. For the SDADA ytem programming oftware, tandard uch a IEC can be ued, which i the international tandard for controller programming language. It pecifie the yntax, emantic and diplay for the PLC programming language. An open tandard communication protocol uch a Ditributed Network Protocol (DNP) can be ued. DNP i a et of tandard and interoperable communication protocol ued between companie in proceing automation ytem. Thee open ource tandard and interoperable platform will allow the ytem deign to be implemented in any of the indutrial proven commercial SCADA ytem. The functionality and proce automation of the ytem can be decribed in two cenario. In the firt cenario, the machine can be operated in a emi-automated mode with online

21 180 Global Navigation Satellite Sytem Signal, Theory and Application telemetric control. The real-time GNSS poitioning i computed and a Geographic Information Sytem (GIS) i utilied at a control centre that operate the machine in a telemetric mode. In the econd cenario, the ytem can be fully automated baed on pre-et intruction and automated integration with GIS planning, databae and geo-coded map. In thi cae, the GIS ytem i uploaded in the field machine. Fig. 7. Propoed network RTK GNSSS and SCADA integration Figure 7 illutrate a developed architecture for the firt cenario, where the machine i remotely controlled from a SCADA centre. In thi cae, SCADA monitoring and controlling hardware and oftware can be automated in the field with a built-in computer that ha all required oftware and i placed inide the machine. The SCADA ytem can be integrated with the GIS through an interface where the mimic are truly geo-referenced map repreenting the reality of the field in a full 3D mode. The GNSS and the primary enor are ent through the RTU to the SCADA ytem to update the GIS map with real-time information. The following procedure can be applied (El-Mowafy and Al-Muawa, 2009): 1. The roving GNSS will be mounted on the field machine and end it obervation to the network centre. Poitioning information i computed at the centre for telemetric control of the machine. 2. The control centre operate the SCADA monitoring and controlling hardware and oftware. The field automated proce i fully monitored and controlled in a remote mode. Report, alarm, trend and hitorical record can be retrieved at the centre. 3. SCADA ytem chooe the planned field work baed on the 3D accurate coordinate of the field point and uing GIS.

22 Precie Real-Time Poitioning Uing Network RTK The centre end primary information to the field RTU unit that control the machine. Thee command can be either full machine command or only the main command that tranfer between phae of field operation if detailed information i pre-fed to the RTU. 5. The RTU baed on a preloaded automation proce program will produce the controlling output baed on a et of variable input from the primary enor a well a the input of the poition information received from the centre. 6. Real-time data from the GNSS and the primary enor are ent remotely from the field through the RTU to the SCADA ytem to update it with the actual work progre in a feedback loop. The logical equence of the oft PLC proce program of the RTU i illutrated in Figure 8 a a flow chart. A the figure depict, the program tart with initialiing and teting the ytem availability and health tatu. It read the current poitioning information of the machine a well a the tatu of the input from primary enor. It execute the RTU proce program while monitoring any alarm or malfunction ignal for an emergency top. The RTU event, log and alarm are ent to the remote SCADA centre. The machine automation in thi cenario can be fully automated in the field. In thi cae, the GNSS unit mounted on the field machine receive the correction from the NRTK and continuouly feed the RTU with the real-time poitioning information repreenting the exact location of the machine in the field. The automation ytem i uploaded on a computer fitted in the machine and all control i pre-programmed and work i executed online in a feedback loop to keep up with the pre-et deign. The real-time 3D accurate coordinate of the field point are to be input to a GIS ytem on board, where it output i fued with the PLC program. Work order can be downloaded remotely to the field RTU at any tage if a change of plan i required. The RTU can alo be connected through a modem to a remote monitoring centre with the ability to control and adjut the field proce baed on any new input. The field automated proce can thu be fully monitored and controlled in a remote mode. Report, alarm, trend and hitorical record can be archived at the centre. To invetigate poitioning preciion that can be obtained from network RTK for machine automation, a tet wa performed in Abu Dhabi. The Abu Dhabi network RTK wa ued in thi tet. The network conit of 20 reference tation with eparating ditance between tation ranging between 60 km to 209 km. Several type of NRTK technique can be implemented a per uer choice, including the VRS method, the MAC approach, the FKP, and tandard RTK uing a ingle nearby reference tation. The propoed approach wa teted in the marine mode for one hour where a Leica 1200 GPS ytem dual-frequency receiver wa mounted on a dredger working in a mall iland cloe to Abu Dhabi main iland. The NRTK poitioning ytem wa operating at a ampling rate of 1 Hz. In addition, the data were internally tored for pot-miion proceing to act a a reference for comparion with network RTK reult to ae it accuracy for the application being teted. The rover data were referenced in thi cae to tation ADCC of the Abu Dhabi continuouly operating reference network. The ditance between thi tation and the tet trajectory wa 6 km on average, giving table ambiguity fixing with precie poitioning output. When comparing the two et of poitioning reult (Network RTK and pot-miion proceing) the difference were at the cm range. The average preciion of the determined poition in NRTK mode wa 2.85 cm for the horizontal component and 4.1 cm for the height. Statitic

23 182 Global Navigation Satellite Sytem Signal, Theory and Application of the poitioning difference between the two method are given in Table 4. During teting, availability of NRTK wa higher than 95%. Thee reult how that the NRTK can be uccefully ued for poitioning of field machine. Fig. 8. Flowchart of the RTK GNSS & SCADA/GIS Logic control

24 Precie Real-Time Poitioning Uing Network RTK 183 S.D Maximum (cm) Average (cm) σ E σ N σ h Table 4. Difference between network RTK and pot-miion poitioning in machine automation teting 7.3 Uing network RTK in the airborne mode The network RTK approach i motly ued in tatic or kinematic ground application. In thi ection, the ue of the NRTK approach in the airborne mode i dicued. At preent, poitioning by GNSS i a widely ued technique in the airborne mode for geo-referencing of aerial mapping data and urveillance by Unmanned Aerial Vehicle (UAV). In aviation, it i etimated that from 2015, mot new commercial aircraft will be fitted with GNSS to enhance precie navigation and make it afer (Pedreira, 2009). However, at the moment, GPS i the only approved ytem a a tand-alone aid for non-preciion approache (Radišić, 2010), e.g. a a upplementary navigation ytem and for poitioning in non afety-of-life application. Thi i mainly due to the need to achieve high level of performance in term of integrity, availability and reliability in the airborne navigation, which GPS on it own cannot reach due to the limited number of atellite available in one ite at any particular intance. Thi ituation i expected to improve with the addition of the new ytem uch a Galileo and Compa. When uing network RTK in the airborne navigation, additional concern have to be addreed, which include: Due to the high dynamic involved in the airborne navigation, a high update rate of ending the correction i needed compared with the rate implemented for land application. Thi rate ha a direct impact on the Time-To-Firt-Fix of phae ambiguitie, and thu on the overall poitioning feaibility and accuracy (El-Mowafy, 2004). The format of GPS meaurement correction hould be tandardied to enure that the ytem i independent of any ingle receiver manufacturer. The ue of the RTCM Verion 3.1 tandard i thu recommended. The main advantage of uing network RTK for precie airborne poitioning can be ummaried a follow: No dedicated ground reference tation are needed for pot-miion or real-time application. Unlike tandard differential poitioning, the ditance between the aircraft receiver and the nearet reference tation doe not preent a concern a long a the aircraft flie within the network RTK area of coverage. In navigation, due to the fact that network RTK uually have an area of coverage that extend to everal hundred of kilometre, each network can cover more than one airport, including mall airport, unlike the current Local Area Augmentation ytem

25 184 Global Navigation Satellite Sytem Signal, Theory and Application (LAAS) implemented only in ome major airport. NRTK ytem can alo be ued in earch and recue operation, emergency landing, road traffic monitoring from the air, a well a emergency repone. Compared to LAAS, no ignificant additional infratructure cot i involved a the hardware and oftware of the GNSS-NRTK are available in mot developed countrie and the etablihment of new network i currently underway or planned in different region worldwide. Network RTK provide cm to decimetre poitioning accuracy even in the cae of malfunctioning of ome reference tation, particularly for dene network. Network RTK can give better runway utiliation by improving airport urface navigation. It can alo enhance air traffic management by increaing dynamic flight planning. The ue of the VRS technique in the airborne mode i not generally recommended ince in thi high velocity environment continuouly updated approximate coordinate have to be ued for the VRS computation. Thi i imilar to having a moving reference tation. A ytem reet hould thu be frequently performed when the VRS coordinate are changing, which will reult in frequent initialiation of the carrier-phae ambiguitie. Therefore, it i preferable to keep the VRS location for the longet poible range. An alternative approach would be to apply the PRS technique, where the PRS point are choen along the path of the final approach and cloe to and at the airport. Furthermore, the duplex communication mode ued in the VRS technique i limited by the ability of the proceing centre to imultaneouly perform calculation for all uer. A thi number grow, extended latency in receiving the correction may reult. Additionally, the poibility of ignal break in the duplex communication mode i more than the cae of uing a one-direction communication. Thu, the ue of a one-directional communication method, e.g. applying the FKP method, would be more appropriate for the airborne mode. The PRS and Mac technique can alo be implemented in the one-directional mode, whereby the PRS or the Mater-Auxiliary tation are elected to cover a pecific area, uch a the airport. The etablihment of ground tranmitter at the airport can improve availability of the correction. The feaibility of uing real-time reference network for poitioning in the airborne mode wa examined uing the DVRS NRTK over the city of Dubai. Flight tet uing a helicopter and a mall fixed-wing airplane were carried out. The trajectory of the fixed-wing aircraft tet i illutrated in Figure 9. The main parameter under invetigation were the achievable accuracy and availability of VRS meaurement. In thee tet, aircraft poition were determined uing a dual-frequency GPS receiver (Leica SR530). The data were proceed in real time at one-econd interval. The DVRS reference tation collect and proce data at five-econd interval. Thu, the NRTK data were interpolated in time for the rover receiver to compute poition at the one-econd interval. To ae performance of NRTK approach for thi tet, the reult were compared with poition determined from a tandard double-difference technique whereby the obervation of the aircraft receiver were tored and proceed in a pot-miion mode. The aircraft data in thi cae were referenced to one of the DVRS network tation located within a range of a few kilometre from the flight route. Precie IGS orbit were ued in the potmiion proceing. The difference between the two method (NRTK and pot-miion proceing) are given in Figure 10. For the tet at hand, the DVRS data were lot for ome period, which ranged from a few econd to three minute. The period when the DVRS

26 Precie Real-Time Poitioning Uing Network RTK 185 data were available are hown in the dahed area in Figure 10. The temporary break in reception of the NRTK data can be attributed to the ue of GSM ignal a the mean of communication between the DVRS centre and the aircraft at the time of the tet. However, the GSM ignal were only ued for teting purpoe. In practice, the problem of break in receiving the network correction can be ignificantly alleviated by uing more robut mean of delivering NRTK ervice to the aircraft Latitude (deg.) Fig. 9. Trajectory of a fixed-wing aircraft tet Longitude (deg.) When comparing the reult of poitioning obtained by the DVRS NRTK with the potmiion double-difference poitioning for the period where NRTK data were received and phae ambiguitie were fixed, the average 2-D and height poitioning dicrepancie between the two method were at a few cm level, a they were 1.6 cm and 2.8 cm repectively. The difference can mainly be attributed to the model aumption and procedure in the two technique. During the period when phae ambiguitie were only olved in a float olution, the difference were 26.3 cm and 52.5 cm. However, when the DVRS NRTK data were lot, poitioning accuracy deteriorated to the metre level. In thi cae, upporting method, uch a good prediction algorithm and integration with other enor, e.g. a geodetic-grade inertial ytem, are needed to cover the hort period when break in reception of the meaurement correction take place. Several method for prediction of NRTK obervation correction a a time erie were invetigated in El-Mowafy, Different time-erie prediction method were invetigated for different type of error. The double exponential moothing prediction approach performed bet in mot of the cae when tudying the atellite clock error correction. Winter' method and the Autoregreive Integrated Moving Average (ARIMA) model were the bet method for predicting the orbital and wet tropopheric error, repectively.

27 186 Global Navigation Satellite Sytem Signal, Theory and Application 250 Flying Height (m) Hz. Error (m) Height Error (m) Time (Sec.) Fig. 10. Fixed-wing aircraft tet reult 8. Reference Al-Shaery, A.M., Lim, S., & Rizo, C. (2010). Functional model of ordinary kriging for medium range real-time kinematic poitioning baed on the Virtual Reference Station technique, Proceeding of 23 rd Int. Tech. Meeting of the Satellite Diviion of the U.S. Int. of Navigation,, pp , Portland, Oregan, USA, September 21-24, Aponte, J., Meng, X., Hill, C., Moore, T., Burbidge M. & Dodon A. (2009). Quality aement of a network-baed RTK GPS ervice in the UK. Journal of Applied Geodey, Vol. 3, pp BKG (2011). Networked Tranport of RTCM via Internet Protocol , Available: Cruddace, P., Wilon, I., Greave, M., Euler, H-J., Keenan, R. & Wüebbena, G. (2002). The Long Road to Etablihing A National Network RTK Solution, Proceeding FIG XXII Int. Congre, Wahington, D.C., April 19-26, 2002.

28 Precie Real-Time Poitioning Uing Network RTK 187 Dai, L., Han, S., Wang, J. & Rizo, C. (2001). A tudy on GPS/GLONASS multiple reference tation technique for precie real-time carrier phae poitioning, Proceeding 14 th Int. Tech. Meeting of the Satellite Diviion of the U.S. Int. of Navigation, pp , Salt Lake City, UT, September 11-14, El-Mowafy, A. (2000). Performance Analyi of the RTK Technique in an Urban Environment, the Autralian Surveyor, Vol. 45, No. 1, pp El-Mowafy, A., Fahir, H., Al Marzooqi, Y., Al Habbai, A. & Babiker, T. (2003). Teting of the DVRS National GPS-RTK Network, Proceeding of the 8 th ISU International Sympoium, Strabourg, France, May 25-28, El-Mowafy, A. (2004). Uing Multiple Reference Station GPS Network for Aircraft Preciion Approach and Airport Surface Navigation, Proceeding of GNSS 2004, The 2004 International Sympoium on GNSS/GPS, Sydney, Autralia, December 6 8, El-Mowafy, A. (2005). Analyi of the Deign Parameter of Multi-Reference Station RTK GPS Network, Journal of Satellite and Land Information Science (SaLIS), Vol. 65, No. 1, pp El-Mowafy, A. (2008). Improving the Performance of RTK-GPS Reference Network for Precie Airborne Navigation, Navigation, Journal of the Intitute Of Navigation (ION), Vol. 57, No. 3, pp El-Mowafy, A. & Al-Muawa, M. (2009). Utilization of GIS and RTK GPS Reference Network for Machine Automation, Proceeding of the 6 th International Sympoium on Mechatronic and it Application, Sharjah, UAE, March 24-26, Euler, H.J., Townend, B.R. & Wübbena, G. (2002). Comparion of Different Propoal for Reference Station Network Information Ditribution Format, Proceeding of the International Technical Meeting, ION GPS-02, pp , Portland, Oregon. September Euler, H-J., Seeger, S., Zelzer, O., Takac, F. & Zebhauer, B. E. (2004). Improvement of Poitioning Performance Uing Standardized Network RTK Meage, Proceeding of ION NTM, San Diego, CA, January 26-28, Fotopoulo, G. (2000). Parameterization of DGPS Carrier Phae Error Over a Regional Network of Reference Station. Mater thei. Department of Geomatic Engineering, Univerity of Calgary, Calgary, Canada. Fotopoulo, G. & Cannon, M. E. (2001). An Overview of Multi-Reference Station Method for cm-level Poitioning. GPS Solution, Vol. 4, No. 3, pp Forberg, R., Strykowki, G. & Tcherning, C. C. (2001). Geoid Model for Dubai Emirate, Report No. SP296, Dubai Municipality. Hieha, C.H. & Wu, J. (2008). Multipath Reduction on Repetition in Time Serie from the Permanent GPS Phae Reidual, The International Archive of the Photogrammetry, Remote Sening and Spatial Information Science, Vol. XXXVII, Part B4, Beijing, pp Hu, G. R., Khoo, V. H. S., Goh, P. C. and Law, C. L. (2002). Internet-baed GPS VRS RTK poitioning with a multiple reference tation network. Journal of Global Poitioning Sytem. Vol. 1, No. 2, pp Hu, G.R., Khoo, H.S., Goh, P.C. & Law, C.L. (2003). Development and Aement of GPS Virtual Reference Station for RTK Poitioning, Journal of Geodey, Vol. 77, No. 5, pp Leica Geo. Sytem (2011). Uing Network RTK, , Available from: LENZ, E. (2004). Networked Tranport of RTCM via Internet Protocol (NTRIP) - Application and Benefit in Modern Surveying Sytem. FIG Working Week 2004, Athen, Greece. May

29 188 Global Navigation Satellite Sytem Signal, Theory and Application Li, X., Zhang, Z. & M. Ge (2011). Regional reference network augmented precie point poitioning for intantaneou ambiguity reolution. Journal of Geodey. Vol. 85, No3, pp Pedreira, P. (2009). Optimitic Outlook for Galileo, GIM International, pp Radišić, T., Novak, D. & Bucak, T. (2010). The Effect of Terrain Mak on RAIM Availability, Journal of Navigation, Vol. 63, No. 1, pp Petrovki, I., Kawaguchi, S., Torimoto, H., Fujii, K., Saano, K., Cannon, M.E. & Lachapelle, G. (2001). Practical Iue of Virtual Reference Station Implementation for Nationwide RTK Network, Proceeding of GNSS 2001, The 5 th GNSS International Sympoium, Seville, Spain, 8-11 May, Takac, F. & Lienhart, W. (2008). SmartRTK: A Novel Method of Proceing Standardied RTCM Network RTK Information for High Preciion Poitioning, Proceeding of ENC GNSS 2008, Touloue, France, April 22-25, Teunien, P. J. G. (1995). The leat quare ambiguity decorrelation adjutment: a method for fat GPS integer ambiguity etimation, Journal of Geodey, Vol. 70, No. 1-2, pp Teunien, P. J. G. & Verhagen, S., (2009). The GNSS ambiguity ratio-tet reviited, Survey Review, Vol. 41, No. 312, pp Teunien, P., Odijk, D.J.G & Zhang, B. (2010). Reult of CORS Network Baed PPP with Integer Ambiguity Reolution, Journal of Aeronautic, Atronautic and Aviation, Serie A, Vol. 42, No. 4, pp Varner, C. (2000). DGPS carrier phae network and partial derivative algorithm. Ph.D Thei, Dept. of Geomatic Engineering, Univerity of Calgary, Calgary, Canada. Vollath, U., Buecherl, A., Landau, H., Pagel, C. & Wager, B. (2000). Multi-bae RTK poitioning uing virtual reference tation. Proceeding 13 th Int Tech Meeting Satellite Diviion, US ION, Salt Lake City, UT, September, Wu, S., Zhang, K., & Silcock D. (2009). Difference in Accuracie and Fitting Surface Plane of Two Error Model for NRTK in GPSnet. Journal of Global Poitioning Sytem, Vol.8, No.2, pp Wu, S., (2009). Performance of Regional Atmopheric Error Model for NRTK in GPSnet and the Implementation of NRTK Sytem, Ph.D Thei, School of Mathematical and Geopatial Science, RMIT Univerity, Melbourne, Autralia. Wübbena, G., Bagge, A. & Schmitz, M. (2001). Network-Baed Technique for RTK Application. Proceeding the GPS JIN 2001 Sympoium, GPS Society, Japan Intitute of Navigation, Tokyo, Japan, November 14-16, Wübbena, G. & Willgali, S. (2001). State Space Approach for Precie Real Time Poitioning in GPS Reference Network. Proceeding of International Sympoium on Kinematic Sytem in Geodey, Geomatic and Navigation, KIS-01, Banff, Canada, June 5-8, Wübbena, G., Schmitz, M. & Bagge, A. (2005). PPP-RTK: Precie Point Poitioning Uing State-Space Repreentation in RTK Network, Proceeding of the 18 th International Technical Meeting of the Satellite Diviion of The Intitute of Navigation ION GNSS 2005, Long Beach, California, September 13-16, 2005, pp Wübbena, G. & Bagge, A. (2006). RTCM Meage Type 59 - FKP for tranmiion of FKP, Verion 1.1, Geo++ GmbH White Paper Nr , , Available: Zebhauer, B.E, Euler, H-J, Keenan, C.R & Wübbena, G. (2002). A Novel Approach for the Ue of Information from Reference Station Network Conforming to RTCM V2.3 and Future V3.0. Proceeding of PLANS 2002, Palm Spring, California, April 15-18, 2002.

30 Global Navigation Satellite Sytem: Signal, Theory and Application Edited by Prof. Shuanggen Jin ISBN Hard cover, 426 page Publiher InTech Publihed online 03, February, 2012 Publihed in print edition February, 2012 Global Navigation Satellite Sytem (GNSS) play a key role in high preciion navigation, poitioning, timing, and cientific quetion related to precie poitioning. Thi i a highly precie, continuou, all-weather, and real-time technique. The book i devoted to preenting recent reult and development in GNSS theory, ytem, ignal, receiver, method, and error ource, uch a multipath effect and atmopheric delay. Furthermore, varied GNSS application are demontrated and evaluated in hybrid poitioning, multi-enor integration, height ytem, Network Real Time Kinematic (NRTK), wheeled robot, and tatu and engineering urveying. Thi book provide a good reference for GNSS deigner, engineer, and cientit, a well a the uer market. How to reference In order to correctly reference thi cholarly work, feel free to copy and pate the following: Ahmed El-Mowafy (2012). Precie Real-Time Poitioning Uing Network RTK, Global Navigation Satellite Sytem: Signal, Theory and Application, Prof. Shuanggen Jin (Ed.), ISBN: , InTech, Available from: InTech Europe Univerity Campu STeP Ri Slavka Krautzeka 83/A Rijeka, Croatia Phone: +385 (51) Fax: +385 (51) InTech China Unit 405, Office Block, Hotel Equatorial Shanghai No.65, Yan An Road (Wet), Shanghai, , China Phone: Fax: