Chapter 1 DIFFERENTIAL GPS

Chapter 1 DIFFERENTIAL GPS 1.1 INTRODUCTION Satellte navgaton systems can provde far hgher accuracy than any other current long and medum range navgaton system. Specfcally, n the case of GPS, dfferental technques have been developed whch can provde accuraces comparable wth current landng systems. The am of ths chapter s to provde an overvew of current DGPS technques and flght applcatons. Due to the exstence of a copous lterature on GPS basc prncples and applcatons, they wll not be deeply covered n ths dssertaton. Only a bref revew of GPS fundamental characterstcs s presented n Annex A, wth an emphass on aspects relevant to the scope of ths dssertaton. Dfferental GPS (DGPS) was developed to meet the needs of postonng and dstance-measurng applcatons that requred hgher accuraces than stand-alone Precse Postonng Servce (PPS) or Standard Postonng servce (SPS) GPS could delver. DGPS nvolves the use of a control or reference recever at a known locaton to measure the systematc GPS errors; and, by takng advantage of the spatal correlaton of the errors, the errors can then be removed from the measurement taken by movng or remote recevers located n the same general vcnty. There have been a wde varety of mplementatons descrbed for affectng such a DGPS system. It s the ntent n ths chapter to characterse varous DGPS systems and compare ther strengths and weaknesses n flght applcatons. Two general categores of dfferental GPS systems can be dentfed: those that rely prmarly upon the code measurements and those that rely prmarly upon the carrer phase measurements. Usng carrer phase, hgh accuracy can be obtaned (centmetre level), but the soluton suffers from nteger ambguty and cycle slps. Whenever a cycle slp occurs, t must be corrected for, and the nteger ambguty must be re-calculated. The pseudorange soluton s more robust, but less accurate (2 to 5 m). It does not suffer from cycle slps and therefore there s no need for re-ntalsaton. 1.2 DGPS CONCEPT A typcal DGPS archtecture s shown n Fgure 1-1. The system conssts of a Reference Recever (RR) located at a known locaton that has been prevously surveyed, and one or more DGPS User Recevers (UR). The RR antenna, dfferental correcton processng system, and datalnk equpment (f used) are collectvely called the Reference Staton (RS). oth the UR and the RR data can be collected and stored for later processng, or sent to the desred locaton n real tme va the datalnk. DGPS s based on the prncple that recevers n the same vcnty wll smultaneously experence common errors on a partcular satellte rangng sgnal. In general, the UR (moble recevers) use measurements from the RR to remove the common errors. In order to accomplsh ths, the UR must smultaneously use a subset or the same set of satelltes as the reference staton. The DGPS postonng equatons are formulated so that the common errors cancel. RTO-AG-160-V21 1-1

(D)GPS Recever Data Lnk (D)GPS Reference Recever Surveyed GPS Antenna Data Lnk Correctons DGPS Ground Staton Fgure 1-1: Typcal DGPS Archtecture. The common errors nclude sgnal path delays through the atmosphere, and satellte clock and ephemers errors. For PPS users, the common satellte errors are resdual system errors that are normally present n the PVT (Poston, Velocty, and Tme) soluton. For SPS users, the common satellte errors (typcally affected by larger onospherc propagaton errors than SPS) also ncluded the ntentonally added errors from Selectve Avalablty (SA), whch have been removed wth the current US-DoD polcy. Errors that are unque to each recever, such as recever measurement nose and multpath, cannot be removed wthout addtonal recursve processng (by the reference recever, user recever, or both) to provde an averaged, smoothed, or fltered soluton [1]. Greater recever nose and multpath errors are present n SPS DGPS solutons. Varous DGPS technques are employed dependng on the accuracy desred, where the data processng s to be performed, and whether real-tme results are requred. If real-tme results are requred then a datalnk s also requred. For applcatons wthout a real-tme requrement, the data can be collected and processed later. The accuracy requrements usually dctate whch measurements are used and what algorthms are employed. Under normal condtons, DGPS accuracy s largely ndependent of whether SPS or PPS s beng used (although, as mentoned before, greater recever nose and multpath errors are present n SPS DGPS). When SA was on, real-tme PPS DGPS had a lower data rate than SPS DGPS because the rate of change of the nomnal system errors was slower than the rate of change of SA. In any case, the user and the Reference Staton must be usng the same servce (ether PPS or SPS). The clock and frequency bases for a partcular satellte wll appear the same to all users snce these parameters are unaffected by sgnal propagaton or dstance from the satellte. The pseudorange and deltarange (Doppler) measurements wll be dfferent for dfferent users because they wll be at dfferent locatons and have dfferent relatve veloctes wth respect to the satellte, but the satellte clock and frequency bas wll be common error components of those measurements. The sgnal propagaton delay s truly a common error for recevers n the same locaton, but as the dstance between recevers ncreases, 1-2 RTO-AG-160-V21

ths error gradually de-correlates and becomes ndependent. The satellte ephemers has errors n all three dmensons. Therefore, part of the error wll appear as a common range error and part wll reman a resdual ephemers error. The resdual porton s normally small and ts mpact s small for smlar observaton angles to the satellte. The accepted standard for SPS DGPS was developed by the Rado Techncal Commsson for Martme Servces (RTCM) Specal Commttee-104 [2, 3]. The RTCM developed standards for use of dfferental correctons, and defned the data format to be used between the reference staton and the user. The data nterchange format for NATO PPS DGPS s documented n STANAG 4392. The SPS reversonary mode specfed n STANAG 4392 s compatble wth the RTCM SC-104 standards. The standards are prmarly ntended for real-tme operatonal use and cover a wde range of DGPS measurement types. Most SPS DGPS recevers are compatble wth the RTCM SC-104 dfferental message formats. DGPS standards have also been developed by the Rado Techncal Commsson for Aeronautcs (RTCA) for specal Category-I (CAT-I) precson approach usng range-code dfferental. The standards are contaned n RTCA document DO-217. Ths document s ntended only for lmted use untl an nternatonal standard can be developed for precson approach [4]. 1.3 DGPS IMPLEMENTATION TYPES There are two prmary varatons of the dfferental measurements and equatons. One s based on rangngcode measurements and the other s based on carrer-phase measurements. There are also several ways to mplement the datalnk functon. DGPS systems can be desgned to serve a lmted area from a sngle reference staton, or can use a network of reference statons and specal algorthms to extend the valdty of the DGPS technque over a wde area. The result s that there s a large varety of possble DGPS system mplementatons usng combnatons of these desgn features. 1.3.1 Rangng-Code Dfferental GPS The rangng-code dfferental technque uses the pseudorange measurements of the RS to calculate pseudorange or poston correctons for the UR. The RS calculates pseudorange correctons for each vsble satellte by subtractng the true range determned by the surveyed poston and the known orbt parameters from the measured pseudorange. The UR recever then selects the approprate correcton for each satellte that t s trackng, and subtracts the correcton from the pseudorange that t has measured. The moble recever must only use those satelltes for whch correctons have been receved. If the RS provdes poston correctons rather than pseudorange correctons, the correctons are smply determned by subtractng the measured poston from the surveyed poston. The advantage of usng poston correctons s obvously the smplcty of the calculatons. The dsadvantage s that the reference recever and the user recever must use the exact same set of satelltes. Ths can be accomplshed by coordnatng the choce of satellte between the RR and the UR, or by havng the RS compute a poston correcton for each possble combnaton of satelltes. For these reasons, t s usually more flexble and effcent to provde pseudorange correctons rather than poston correctons. The RTCM SC-104, NATO STANAG 4392, and RTCA DO-217 formats are all based on pseudorange rather than poston correctons. The pseudorange or poston correctons are tme tagged wth the tme that the measurements were taken. In real-tme systems, the rate of change of the correctons s also calculated. Ths allows the user to propagate the correctons to the tme that they are actually appled to the user poston soluton. Ths reduces the mpact of data latency on the accuracy of the system, but does not elmnate t entrely. SPS correctons become fully uncorrelated wth the user measurements after about 2 mnutes. Correctons used after two mnutes may produce solutons whch are less accurate than stand-alone SPS GPS. PPS correctons can reman correlated wth the user measurements for 10 mnutes or more under bengn (slowly changng) onospherc condtons. RTO-AG-160-V21 1-3

There are two ways of pseudorange data processng: post-msson and real-tme processng. The advantage of the post-msson soluton over the real-tme one, s that t s more accurate, because the user can easly detect blunders and analyse the resduals of the soluton. On the other hand the man dsadvantage of the post-msson soluton s that the results are not avalable mmedately for navgaton. The typcal algorthm of the rangng-code DGPS post-processed soluton s the double dfference pseudorange. The mathematcal models for both sngle dfference and double dfference observables are developed n the followng paragraphs. 1.3.1.1 Sngle Dfference etween Recevers Fgure 1.2 shows the possble pseudorange measurements between two recevers (k. l) and two satelltes (p, q). If pseudorange 1 and 2 from Fgure 1.2 are dfferenced, then the satellte clock error and satellte orbt errors wll be removed. Moreover, SA wll be reduced and wll be removed completely only f the sgnals transmtted to each recever, are emtted exactly at the same tme. The resdual error from SA s not a problem for post-processed postonng, where t s easy to ensure that the dfferencng s done between pseudoranges observed at the same tme [5]. Any atmospherc errors wll also be reduced sgnfcantly wth sngle dfferencng. Fgure 1-2: Pseudorange Dfferencng. The basc mathematcal model for sngle dfference pseudorange observaton s the followng (refer to equaton A.7 of Secton A.4 n Annex A): ( ) p p p p p p Pk Pl = ρk ρl dtk dtl c + dk, p dl, p+ dk, p dlp, + εp (1.1) 1-4 RTO-AG-160-V21

where P p s the pseudorange measurement, ρ p denotes the geometrc dstance between the statons and p satellte, dt denotes the recever s clock offsets, d p, denotes the recever s hardware code delays, d p, denotes the multpath of the codes, ε p denotes the measurement nose and c s the velocty of lght. Equaton (1.1) represents the sngle dfference pseudorange observable between recevers. Another type of sngle dfference apart from (1.1), s known as between-satellte sngle dfference. There are four unknowns n equaton (1.1) assumng that the co-ordnates of staton k are known and that the dfference n clock drfts s one unknown. Hence, four satelltes are requred to provde four sngle dfference equatons n order to solve for the unknowns. Sngle dfferences wth code observatons are frequently used n relatve (dfferental) navgaton [6]. 1.3.1.2 Double Dfference Observable Usng all pseudoranges shown n Fgure 1-2, dfferences are formed between recevers and satelltes. Double dfferences are constructed by takng two between-recever sngle dfferences and dfferencng these between two satelltes. Ths procedure removes all satellte dependent, recever dependent and most of the atmospherc errors (f the dstance between the two recevers s not too large). The derved equaton s: j where d p, denotes the total effect of multpath. p q p q p q p q j P P P P = ρ ρ ρ ρ + d, (1.2) k k l l k There are three unknowns n equaton (1.2); the co-ordnates of staton l. A mnmum of four satelltes s requred to form a mnmum of three double dfference equatons n order to solve for the unknowns. Usng the propagaton of errors law, t s shown that the double dfference observables are twce as nosy as the pure pseudoranges [5]: k l 2 2 2 2 σ = σ + σ + σ + σ = 2 σ (1.3) DD P P P P P but they are more accurate, because most of the errors are removed. Note that multpath remans, because t cannot be modelled and t s ndependent for each recever. l p 1.3.2 Carrer-Phase Dfferental GPS The carrer-phase measurement technque uses the dfference between the carrer phases measured at the RR and UR. A double-dfferencng technque s used to remove the satellte and recever clock errors. The frst dfference s the dfference between the phase measurement at the UR and the RR for a sngle satellte. Ths elmnates the satellte clock error whch s common to both measurements. Ths process s then repeated for a second satellte. A second dfference s then formed by subtractng the frst dfference for the frst satellte from the frst dfference for the second satellte. Ths elmnates both recever clock errors whch are common to the frst dfference equatons. Ths process s repeated for two pars of satelltes resultng n three double-dfferenced measurements that can be solved for the dfference between the reference staton and user recever locatons. Ths s nherently a relatve postonng technque, therefore the user recever must know the reference staton locaton to determne ts absolute poston. More detals of these processes are llustrated n the followng subsectons were the varous observaton equatons are presented. RTO-AG-160-V21 1-5

1.3.2.1 Sngle Dfference Observable The sngle dfference s the nstantaneous phase dfference between two recevers and one satellte. It s also possble to defne sngle dfferences between two satelltes and one recever. Usng the basc defnton of carrer-phase observable presented n equaton (A.23) of Annex A, the phase dfference between the two recevers A and, and satellte s gven by: Φ ( τ ) = Φ ( τ ) Φ ( τ ) (1.4) A and can be expressed as: Φ f ( τ ) = ρ ( ) Φ ( ) t τ N + (1.5) c where N = N N. Hence, wth four satelltes, j, k and l: Φ Φ j A f ( τ ) = ρ ( ) Φ ( ) t τ N k f k k k + ; Φ ( ) ( ) Φ ( ) τ = ρ t τ N c + ; c f j j j ( τ ) = ρ ( ) Φ ( ) t τ N l f l l l + ; Φ ( ) ( ) Φ ( ) τ = ρ t τ N c +. c 1.3.2.2 Double Dfference The double dfference s formed from subtractng two sngle dfferences measured to two satelltes and j. The basc double dfference equaton s: j Φ ( τ ) = Φ ( τ ) Φ ( τ ) (1.6) whch smplfes to: j f j j Φ ( τ ) = ρ ( ) t N (1.7) c j where N N j j = N, and the only unknowns beng the double-dfference phase ambguty N the recever co-ordnates. The local clock error s dfferenced out. and Two recevers A and, and four satelltes, j, k, and l, wll gve 3 double dfference equatons wth j unknown co-ordnates (X, Y, Z), of A and, and the unknown nteger ambgutes N k, N, l and N : j f j j Φ ( τ ) = ρ ( ) t N k f k k ; Φ ( ) ( ) c τ = ρ t N ; c l f l l Φ ( τ ) = ρ ( ) t N. c Therefore, the double dfference observaton equaton can be wrtten as [7]: 1-6 RTO-AG-160-V21

where: Φ Φ Φ Φ Φ Φ X dx Y dy Z dz X dx Y dy Z dz A + A + A + + + A A A Φ Φ Φ Φ + N dn + N dn + N dn + C dc + = + v 1 2 1 2 X A, YA, Z A = Co-ordnates of Recever A; X, Y, Z = Co-ordnates of Recever ; N 1, N 2, N 3 = Integer Ambgutes; C = Tropospherc Factor; O C ( Φ Φ ) = Observed mnus Computed Observable; and ν = Resdual. 3 (1.8)...... O C 3 ( Φ Φ ) From equaton (1.8) the unknown recever co-ordnates can be computed. It s necessary, however, to determne the carrer phase nteger ambgutes (.e., the nteger number of complete wavelengths between the recever and satelltes). In certan surveyng applcatons, ths nteger ambguty can be resolved by startng wth the moble recever antenna wthn a wavelength of the reference recever antenna. oth recevers start wth the same nteger ambguty, so the dfference s zero and drops out of the double-dfference equatons. Thereafter, the phase shft that the moble recever observes (whole cycles) s the nteger phase dfference between the two recevers. For other applcatons where t s not practcal to brng the reference and moble antennas together, the reference and moble recevers can solve for the ambgutes ndependently as part of an ntalsaton process. One way s to place the moble recever at a surveyed locaton. In ths case the ntal dfference s not necessarly zero, but t s an easly calculated value. For some applcatons, t s essental to be able to solve for nteger ambguty at an unknown locaton or whle n moton (or both). In ths case, solvng for the nteger ambguty usually conssts of elmnatng ncorrect solutons untl the correct soluton s found. A good ntal estmate of poston (such as from rangng-code dfferental) helps to keep the ntal number of canddate solutons small [8]. Redundant measurements over tme and/or from extra satellte sgnals are used to solate the correct soluton. These search technques can take as lttle as a few seconds or up to several mnutes to perform and can requre sgnfcant computer processng power. Ths verson of the carrer-phase DGPS technque s typcally called Knematc GPS (KGPS). If carrer track or phase lock on a satellte s nterrupted (cycle slp) and the nteger count s lost, then the ntalsaton process must be repeated for that satellte. Causes of cycle slps range from physcal obstructon of the antenna to the sudden acceleraton of the user platform. Output data flow may also be nterrupted f the recever s not collectng redundant measurements form extra satelltes to mantan the poston soluton. If a precse poston soluton s mantaned, re-ntalsaton for the lost satellte can be almost mmedate. Developng a robust and rapd method of ntalsaton and re-ntalsaton s the prmary challenge facng desgners of real-tme systems that have a safety crtcal applcaton such as arcraft precson approach. A descrpton of technques for solvng ambgutes both n real-tme and post-processng applcatons, together wth nformaton about cycle slps repar technques can be found n the references [9 17]. 1.3.3 DGPS Datalnk Implementatons DGPS can also be mplemented n several dfferent ways dependng on the type of datalnk used. The smplest way s no datalnk at all. For non-real-tme applcatons, the measurements can be stored n RTO-AG-160-V21 1-7

the recever or on sutable meda and processed at a later tme. In most cases to acheve surveyng accuraces, the data must be post-processed usng precse ephemers data that s only avalable after the survey data has been collected. Smlarly, for some test applcatons the cost and effort to mantan a realtme datalnk may be unnecessary. Nevertheless, low-precson real-tme outputs can be useful to confrm that a test s progressng properly even f the accuracy of the results wll be enhanced later. Dfferental correctons or measurements can be uplnked n real-tme from the reference staton to the users. Ths s the most common technque where a large number of users must be served n real-tme. For mltary purposes and propretary commercal servces, the uplnk can be encrypted to restrct the use of the DGPS sgnals to a selected group of users. Dfferental correctons can be transmtted to the user at dfferent frequences. Wth the excepton of satellte datalnks there s generally a trade-off between the range of the system and the update rate of the correctons [18, 19]. As an example Table 1-1 lsts a number of frequency bands, the range, and the rate at whch the correctons could be updated usng the standard RTCM SC-104 format [2, 3, 20]. Table 1-1: DGPS Datalnk Frequences Frequency Range (km) Update Rate (sec) LF (30 300 khz) > 700 < 20 MF (300 khz 3 MHz) < 500 5 10 HF ( 3 MHz 25 MHz) < 200 5 VHF (30 MHz 300 MHz) < 100 < 5 L and (1 GHz 2 GHz) Lne of Sght Few Seconds An uplnk can be a separate transmtter/recever system or the DGPS sgnals can be supermposed on a GPS-lnk L-band rangng sgnal. The uplnk acts as a pseudo-satellte or pseudolte and delvers the rangng sgnal and DGPS data va the RF secton of the user recever, much n the same way the GPS navgaton message s transmtted. The advantages are that the addtonal rangng sgnal(s) can ncrease the avalablty of the poston soluton and decrease carrer-phase ntalsaton tme. However, the RS and URs become more complex, and the system has a very short range (a few klometres at the most). Ths s not only because of the lne of sght restrcton, but also the power must be kept low n order to avod nterference wth the real satellte sgnals (.e., the pseudolte can become a GPS jammer f t overpowers the GPS satellte sgnals). A downlnk opton s also possble from the users to the RS or other central collecton pont. In ths case the dfferental solutons are all calculated at a central locaton. Ths s often the case for test range applcatons where precse vehcle trackng s desred, but the nformaton s not used aboard the vehcle. The downlnk data can be poston data plus the satellte tracked, or pseudorange and deltarange measurements, or t can be the raw GPS sgnals translated to an ntermedate frequency. The translator method can often be the least expensve wth respect to user equpment, and therefore s often used n muntons testng where the user equpment may be expendable. More detals about these applcatons are gven n Chapter 4. 1.3.4 Local Area and Wde Area DGPS The accuracy of a DGPS soluton developed usng a sngle RS wll degrade wth dstance from the RS ste. Ths s due to the ncreasng dfference between the reference and the user recever ephemers, 1-8 RTO-AG-160-V21

onospherc, and tropospherc errors. The errors are lkely to reman hghly correlated wthn a dstance of 350 km [20], but practcal systems are often lmted by the datalnk to an effectve range of around 170 km. Such systems are usually called Local Area DGPS (LADGPS). DGPS systems that compensate for accuracy degradaton over large areas are referred to as wde area DGPS (WADGPS) systems. They usually employ a network of reference recevers that are coordnated to provde DGPS data that s vald over a wde coverage area. Such systems typcally are desgned to broadcast the DGPS data va satellte, although a network of ground transmsson stes s also feasble. A user recever typcally must employ specal algorthms to derve the onospherc and tropospherc correctons that are approprate for ts locaton from the observatons taken at the varous reference stes. The Unted States, Canada, Europe, Japan, and Australa have developed or are plannng to deploy WADGPS systems transmttng from geostatonary satelltes for use by commercal avaton [21]. The satelltes can also provde GPS-lke rangng sgnals. Other natons may partcpate by provdng clock correctons only from sngle stes or small networks, requrng the user to derve onospherc correctons from an onospherc model or dual-frequency measurements. Some commercal DGPS servces broadcast the data from multple reference statons va satellte. However, several such systems reman a group of LADGPS rather than WADGPS systems. Ths s because the reference statons are not ntegrated nto a network, therefore the user accuracy degrades wth dstance from the ndvdual reference stes. 1.4 DGPS ACCURACY Controlled tests and recent extensve operatonal use of DGPS, have repeatedly demonstrated that DGPS (pseudorange) results n an accuracy of the order of about 10 metres. Ths fgure s largely rrespectve of recever type, whether or not SA s n use, and over dstances of up to 500 km from the Reference Staton [23, 24]. Wth KGPS postonng systems, requrng the resoluton of the carrer phase nteger ambgutes whlst on the move, centmetre level accuracy can be acheved [9, 25]. Many recent applcatons of DGPS use C/A code pseudorange as the only observable, wth acheved accuraces of 1 to 5 m n real-tme. Other applcatons use both pseudorange (C/A or P code) and carrer phase observables. Very Precse DGPS (VPDGPS) and Ultra Precse DGPS (UPDGPS) are the state-ofthe-art ASHTECH packages, takng advantage of precse dual band P code pseudorange and carrer phase observables and s capable of On-The-Fly (OTF) ambguty resoluton. ASHTECH has developed varous technques whch acheve ncreased accuracy at the expense of ncreased complexty (many other recever manufacturers delver lkewse solutons). The ASHTECH classfcaton scheme of these technques s presented n Table 1-2 [26]. Table 1-2: ASHTECH Classfcaton Scheme of DGPS Technques Name Descrpton RMS DGPS C/A code pseudorange 1 5 m PDGPS P code pseudorange 0.1 1 m VPDGPS Addton of dual band carrer phase 5 30 cm UPDGPS Above wth nteger ambgutes resolved < 2 cm A dscusson of DGPS error sources s presented below, together wth a comparson between nondfferental GPS and DGPS error budgets. RTO-AG-160-V21 1-9

1.5 DGPS ERROR SOURCES The major sources of error affectng stand-alone GPS (see Annex A) are the followng: Ephemers Error; Ionospherc Propagaton Delay; Tropospherc Propagaton Delay; Satellte Clock Drft; Multpath; Recever Nose and clock drft; and Selectve Avalablty Errors (only SPS applcatons). Table 1-3 summarses the above stated error sources gvng an estmaton of ther magntudes and the possble mprovement provded by DGPS [18]. Table 1-3: Error Sources n DGPS Error Source Stand Alone (m) DGPS (m) Ephemers 5 20 0 1 Ionosphere 15 20 2 3 Troposphere 3 4 1 Satellte Clock 3 0 Multpath 2 2 Recever Nose 2 2 Selectve Avalablty 50 0 It should be stated that the error from multpath s ste dependent and the value n Table 1-3 s only an example. The recever clock drft s not mentoned n Table 1-3, because t s usually treated as an extra parameter and corrected n the standard soluton. Furthermore, t does not sgnfcantly add to dfferental errors. Multpath and recever nose errors cannot be corrected by DGPS. The strategy used for correctng GPS errors and nduced bases s the followng: Selectve Avalablty Errors. These errors are only of concern to the SPS user. They resemble the naturally occurrng ephemers and clock errors, except that they can be larger n magntude and can change more rapdly. The epslon error can be a three dmensonal error. Therefore, part of the error wll appear as a common range error and part wll reman a resdual ephemers error. The resdual porton s normally small and ts mpact remans small for smlar look angles to the satellte. The dther error can appear as a tme and frequency bas. Ths wll be an error common to all recevers and wll not be affected by sgnal propagaton or dstance from the satellte. However, snce t s rapdly changng, any delay between the tme of measurement at the reference staton and tme of use at the user recever wll result n a resdual clock error. 1-10 RTO-AG-160-V21

SPS DGPS systems are normally desgned wth a rate-of-change term n the correctons and rapd update rates to mnmse ths effect. Ionospherc and Tropospherc Delays. For users near the reference staton, the respectve sgnal paths to the satellte are close enough together that the compensaton s almost complete. As the user to RS separaton s ncreased, the dfferent onospherc and tropospherc paths to the satelltes can be far enough apart that the onospherc and tropospherc delays are no longer common errors. Thus, as the dstance between the RS and user recever ncreases the effectveness of the atmospherc delay correctons decreases. Ephemers Error. Ths error s effectvely compensated unless t has qute a large out-of-range component (e.g., 1000 metres or more due to an error n a satellte navgaton message). Even then, the error wll be small f the dstance between the reference recever and user recever s small. Satellte Clock Error. Except n a satellte falure stuaton, ths error s more slowly changng than the SA dther error. For all practcal purposes, ths error s completely compensated, as long as both reference and user recevers employ the same satellte clock correcton data. Table 1-4 shows the error budget determned for a SPS DGPS system wth ncreasng dstances from the Reference Staton. Table 1-4: SPS DGPS Errors (ft) wth Increasng Dstance from the Reference Staton ERROR SOURCES 0 NM 100 NM 500 NM 1000 NM Space Segment: Clock Errors 0 0 0 0 Control Segment: Ephemers Errors 0 0.3 1.5 3 SA 0 0 0 0 Propagaton Errors: Ionosphere 0 7.2 16 21 Troposphere 0 6 6 6 TOTAL (RMS) 0 9.4 17 22 User Segment: Recever Nose 3 3 3 3 Multpath 0 0 0 0 UERE (RMS) 3 9.8 17.4 22.2 As already mentoned, the correlaton of the errors experenced at the RS and the user locaton s largely dependent on the dstance between them. As the separaton of the user from the RS ncreases so does the probablty of sgnfcant dfferng onospherc and tropospherc condtons at the two stes. Smlarly, the ncreasng separaton also means that a dfferent geometrcal component of the ephemers error s seen RTO-AG-160-V21 1-11

by the RR and UR. Ths s commonly referred to as Spatal Decorrelaton of the ephemers and atmospherc errors. In general, the errors are hghly correlated for a user wthn 350 km of the RS. In most cases however, f the dstance s greater than 250 km the user wll obtan better results usng correcton models for onospherc and tropospherc delay [18, 27]. Snce the RR nose and multpath errors are ncluded n the dfferental correctons and become part of the user s error budget (root-sum-squared wth the user recever nose and multpath errors), the recever nose and multpath error components n the non-dfferental recever can be lower than the correspondent error components experenced n the DGPS mplementaton. The other type of error ntroduced n real-tme DGPS postonng systems s the datalnk age of correctons. Ths error s ntroduced due to the latency of the transmtted correctons (.e., the transmtted correctons of epoch t 0 arrve at the movng recever at epoch t0 + dt). These correctons are not the correct ones, because they were calculated under dfferent SA/AS condtons. Hence, the co-ordnates of the UR would be slghtly offset. 1.6 INTEGRITY ISSUES FOR AIRCRAFT NAVIGATION At the moment, satellte navgaton systems are only certfed to be used as a supplementary mean of arcraft navgaton. Contrary to the systems n use, GPS s, as yet, only certfable for arcraft navgaton f t s ntegrated wth other navgaton systems. The reason s not the accuracy but ntegrty. Accordng to the US Federal Radonavgaton Plan [28], Integrty s the ablty of a system to provde tmely warnngs to users when the system should not be used for navgaton. Another defnton s: Wth probablty P, ether the horzontal radal poston error does not exceed a pre-specfed threshold R, or an alarm s rased wthn a tme-to alarm nterval of duraton T when the horzontal radal poston error exceeds a prespecfed threshold R. To detect that the error s exceedng a threshold, a montor functon has to be nstalled wthn the navgaton system. Ths s also the case wthn the GPS system n the form of the ground segment. However, for ths system, the tme to alarm (TTA) s n the order of several hours, that s even too long for the cruse where a TTA of 60 seconds s requred (an autoland system for zero meter vertcal vsblty must not exceed a TTA of 2 seconds). Varous methods have been proposed and practcally mplemented for stand-alone GPS ntegrty montorng. A growng famly of such mplementatons, already very popular n avaton applcatons, ncludes the so called Recever Autonomous Integrty Montorng (RAIM) technques. Detals about RAIM technques can be found n the references [4, 22]. Regardng DGPS, t should be underlned that t does more than ncreasng the GPS postonng accuracy, t also enhances GPS ntegrty by compensatng for anomales n the satellte rangng sgnals and navgaton data message. The range and range rate correctons provded n the rangng-code DGPS correcton message can compensate for ramp and step type anomales n the ndvdual satellte sgnals, untl the correctons exceed the maxmum values or rates allowed n the correcton format. If these lmts are exceeded, the user can be warned not to use a partcular satellte by placng do-not-use bt patterns n the correctons for that satellte (as defned n STANAG 4392 or RTCM SC-104 message formats) or by omttng the correctons for that satellte. As mentoned before, step anomales wll normally cause carrerphase DGPS recevers to lose lock on the carrer phase, causng the reference and user recevers to rentalse. UR nose, processng anomales, and multpath at the user GPS antenna cannot be corrected by a DGPS system. These errors are ncluded n the overall DGPS error budget. Errors n determnng or transmttng the satellte correctons may be passed on to the dfferental user f ntegrty checks are not provded wthn the RS. These errors can nclude naccuraces n the RS antenna locaton that bas the correctons, systematc multpath due to poor antenna sghtng (usually n low elevaton angle satelltes), algorthmc errors, recever nter-channel bas errors, recever clock errors, and communcaton errors. For these reasons, typcal WADGPS and LADGPS RS desgns also nclude ntegrty checkng provsons to guarantee the valdty of the correctons before and after broadcast [21, 29]. 1-12 RTO-AG-160-V21

1.7 DGPS AUGMENTATION SYSTEMS Varous strateges have been developed for ncreasng the levels of ntegrty, accuracy and avalablty of DGPS-based navgaton/landng systems. These nclude both Space-ased Augmentaton Systems (SAS) and Ground-ased Augmentaton Systems (GAS). Partcularly, the Amercan Wde Area Augmentaton System (WAAS) and the European Geostatonary Navgaton Overlay System (EGNOS) are examples of SAS. In these systems, geostatonary satelltes (INMARSAT-3) are used to broadcast varous sgnals, computed through a ground network of Integrty Montorng Statons and transmtted from a dedcated Earth Staton. In the case of WAAS (Fgure 1-3), the geostatonary (GEO) satelltes broadcast the followng [30]: GPS Use/Don t Use Warnng (Integrty Sgnals); Correctons for each SV: clock, ephemers, onospherc (to ncrease Accuracy); and Rangng Sgnals (to ncrease Avalablty). INMARSAT-3 GPS True Locaton Space Segment Indcated Locaton Atmospherc Effects Integrty Processng Vector Correctons Use/Don t Use Rangng Sgnal User Segment Navgaton Earth Staton (Prmary/Standby) Integrty Montorng Statons Ground Segment Fgure 1-3: Wde Area Augmentaton System. WAAS s desgned to provde precson approach capablty (3-dmensonal gudance) for Category 1 (CAT-1) approaches wth the followng avalablty: etter Than 95% Avalable n the majorty of Contnental US (CONUS); and Rest of U.S. Avalable, but less than 95%. Furthermore, for en-route through non-precson approaches the followng avalablty s specfed: 50% of Contnental US etter than 99.9% Avalablty; and Rest of U.S. Avalable, but less than 99.9%. The Vertcal Protecton Level (VPL) currently offered by the WAAS servce n CONUS s shown n Fgure 1-4. RTO-AG-160-V21 1-13

Fgure 1-4: WAAS Vertcal Protecton Level. The objectve of LAAS (Fgure 1-5) s to provde category II and III Precson Approach (PA) at those arports that requre the capablty and CAT-I PA at those facltes where WAAS PA s not avalable. True Locaton GPS Indcated Locaton Atmospherc Effects Pseudolte 50 NM User Segment Rangng Sgnal Scalar Correctons (each SV) Use/Don t Use Reference Staton Fgure 1-5: Local Area Augmentaton System. 1-14 RTO-AG-160-V21

For ths purpose, the Local Reference Staton broadcasts: GPS Use/Don t Use Warnng (to ncrease Integrty); and Scalar Correctons (to ncrease Accuracy). Local Pseudoltes roadcastng (LP) s mplemented n order to make avalable addtonal rangng sgnals for an ncreased avalablty and accuracy [31]. More detaled nformaton about recent LAAS, WAAS and Pseudoltes systems developments can be found n the references [30 33]. 1.8 REFERENCES [1] Chao, C.H. (1998). Hgh Precson Dfferental GPS. MSc Dssertaton. Insttute of Engneerng Surveyng and Space Geodesy (Insttute of Engneerng Surveyng and Space Geodesy (IESSG)) Unversty of Nottngham. [2] RTCM Specal Commttee No. 104. (1990). RTCM Recommended Standards for Dfferental NAVSTAR GPS Servce. Rado Techncal Commttee for Martme Servces. Paper 134-89/SC104-68. Washngton DC (USA). [3] RTCM Specal Commttee No. 104. (1994). RTCM Recommended Standards for Dfferental NAVSTAR GPS Servce. Rado Techncal Commttee for Martme Servces. Paper 194-93/SC104- STD. Washngton DC (USA). [4] Jont Program Offce (JPO). (1997). NAVSTAR GPS User Equpment, Introducton. Publc Release Verson. US Ar Force Space Systems Dvson, NAVSTAR-GPS Jont Program Offce (JPO). Los Angeles AF, Calforna (USA). [5] Walsh, D. (1994). Knematc GPS Ambguty Resoluton. PhD Thess, Insttute of Engneerng Surveyng and Space Geodesy (IESSG), Unversty of Nottngham. [6] Seeber, G. (1994). Satellte Geodesy. Second Edton. Artech House Publshers. New York (USA). [7] Ashkenaz, V. (1997). Prncples of GPS and Observables. Lecture Notes, Insttute of Engneerng Surveyng and Space Geodesy (IESSG) Unversty of Nottngham. [8] Ashkenaz, V., Moore, T. and Westrop, J.M. (1990). Combnng Pseudo-range and Phase for Dynamc GPS. Paper presented at the Internatonal Symposum on Knematc Systems n Geodesy, Surveyng and Remote Sensng. London (UK). [9] Ashkenaz, V., Foulkes-Jones, G.H., Moore, T. and Walsh, D. (1993). Real-tme Navgaton to Centmetre Level. Paper presented at DSNS93, the 2nd Internatonal Symposum on Dfferental Navgaton. Amsterdam (The Netherlands). [10] Euler, H.J. and Landau, H. (1992). Fast GPS Ambguty Resoluton On-The-Fly for Real-tme Applcatons. Paper presented at the 6th Internatonal Geodetc Symposum on Satellte Postonng. Columbus (Oho). [11] Euler, H.J. (1994). Achevng hgh-accuracy relatve postonng n real-tme: system desgn, performance and real-tme results. Proceedngs of the 4th IEEE Plans Conference. Las Vegas (NV). [12] Hansen, P. (1994). Real-tme GPS Carrer Phase Navgaton. The Unversty of Nottngham, Insttute of Engneerng Surveyng and Space Geodesy (IESSG). Paper presented at the DSNS-94 Conference. London (UK). RTO-AG-160-V21 1-15

[13] Hatch, R.R. (1990). Instantaneous Ambguty Resoluton. Presented at the KIS Symposum. anff (Canada). [14] Hatch, R.R. (1991). Ambguty Resoluton Whle Movng, Expermental Results. Proc. of ION GPS-91, the 4th Internatonal Techncal Meetng of the Satellte Dvson of the US Insttute of Navgaton. Albuquerque (NM). [15] Kleusberg, A. (1986). Knematc Relatve Postonng Usng GPS Code and Carrer eat Phase Observatons. 2nd Marne Geodesy Symposum. London (UK). [16] Landau, H. (1992). On-The-Fly Ambguty Resoluton Usng Dfferental P-code Group and Phase Delay Measurements. 5th Internatonal Techncal Meetng of the Satellte Dvson of the US Insttute of Navgaton. Orlando (USA). [17] Mader, G. (1986). Dynamc Postonng usng GPS Carrer Phase Measurements. Manuscrpte Geodaetca. Volume 36 (86). [18] Moore, T. (2002). An Introducton to Dfferental GPS. Lecture Notes, Insttute of Engneerng Surveyng and Space Geodesy (IESSG) Unversty of Nottngham (UK). [19] Moore, T. (2002). GPS Orbt Determnaton and Fducal Networks. Lecture Notes, Insttute of Engneerng Surveyng and Space Geodesy (IESSG) Unversty of Nottngham (UK). [20] Cross, P.A. and Roberts, W.D.S. (1990). Dfferental Offshore Postonng usng lock II GPS Satelltes. 7th Internatonal Symposum of the Hydrographc Socety (Hydro 90). Southampton (UK). [21] Keth, A. (2000). Usng Wde Area Dfferental GPS to mprove total system error for precson flght Operatons. PhD Thess. Stanford Unversty (USA). [22] Parknson,.W., Splker, J.J., Jr. Edtors. (1996). Global Postonng System: Theory and Applcatons Volume II. Progress n Astronautcs and Aeronautcs. Vol. 163. Publshed by the Amercan Insttute of Aeronautcs and Astronautcs. [23] Moore, T. (2002). Other Satellte Navgaton Systems. Lecture Notes, Insttute of Engneerng Surveyng and Space Geodesy (IESSG) Unversty of Nottngham (UK). [24] Yuan, J., Gu, X., Jacob, T. and Schanzer, G. (1990). Error Correcton for Dfferental GPD wth Long Separated Ground Statons and User for Arcraft Landng. Insttute of Gudance and Control. Techncal Unversty raunschweg (Germany). [25] Ashkenaz, V., Summerfeld, P. and Westrop, J. (1990). Knematc Postonng by GPS. Tré a part des Cahers du Centre Européen de Géodynamque et de Sésmologe, Volume 2. [26] Qn, X., Gourevtch, S. and Kuhl, M. (1992). Very Precse Dfferental GPS Development Status and Test Results. Proc. of ION GPS-92, 5th Internatonal Techncal Meetng of the Satellte Dvson of the US Insttute of Navgaton. Albuquerque (USA). [27] Dodson, A.H. (2002). Propagaton Effects on GPS Measurements. Lecture Notes, Insttute of Engneerng Surveyng and Space Geodesy (IESSG) Unversty of Nottngham (UK). [28] US Department of Defence and US Department of Transportaton. (2001). Federal Radonavgaton Plan 2001. Natonal Techncal Informaton Servces. Sprngfeld (USA). Doc. DOT-VNTSC- RSPA-01-3/DOD-4650.5. 1-16 RTO-AG-160-V21

[29] Ko, P.Y. (1997). Wde Area Dfferental GPS (WADGPS). PhD Thess. Stanford Unversty (USA). [30] Dye, S. and ayln, F. (2004). The GPS Manual Prncples and Applcatons. Second Edton. ayln Publcatons (USA). [31] Cobb, S.A. (1997). GPS Pseudoltes: Theory, Desgn and Applcatons. PhD Dssertaton. Stanford Unversty (USA). [32] Chao, Y. (1997). Real Tme Implementaton of Wde Area Augmentaton System for Global Postonng wth an Emphass on Ionospherc Modelng. PhD Thess. Stanford Unversty (USA). [33] Lee, J.S. (2005). GPS-ased Arcraft Landng Systems wth Enhanced Performance eyond Accuracy. PhD Dssertaton. Stanford Unversty (USA). RTO-AG-160-V21 1-17

1-18 RTO-AG-160-V21