CA2757769A1 - Removing biases in dual frequency gnss receivers using sbas - Google Patents

Abstract

A method for removing biases in dual frequency GNSS receivers circumvents the need for ionosphere corrections by using L2(P) in combination with either L1(P) or L1(C/A) to form ionosphere-free ranges. A table of biases is stored in microprocessor controller memory and utilized for computing a location using corrected ionosphere-free pseudo ranges, A system for removing biases in dual frequency GNSS receivers includes a dual frequency GNSS
receiver and a controller microprocessor adapted to store a table of bias values for correcting pseudo ranges determined using L2(P) in combination with either L1(F) or L1(C/A).

Description

CA 02]5]]69 2011-10.05 REMOVING BIASES N DUAL FREQUENCY GNSS

RECEIVERS 2. 9.E NG S13AS

i'ls. ~ l I l I:I E{ 1 1:" `1'0 R1_A_.A4'F',l_) APPLICATION' 100011 '['his application claims priority in U.S. 13atcnt Application No.
12/401.948, filed Maceli l 1.2009, which is inccorporatcd herein by reference 13At: KG ROB ND OF THE iN 1 N"1 I )l 1E? I . Field of tile Invention [130112] The present invention relates generall`' to global navigation satellite systems ( NSS), and in particular to removing biases in dual frequency GNSS receivers.

2. Description of the Related An 15 [0003] The Global Positioning Sy stern (OPS) vas established by the United States euverrlr lent, and employs a constellation of x4 or more satellite in well-=tlefined orbits at an altitude ot'appro inn4at:ely 6.500 km. I here satellites continually transmit microwave 1_:--hand radio signals in two frequency bands. centered at 1573.42 %1411-1z. and 1227A 4111m, denoted as 1.I and l_2 re zpec,tivel ,. "1'hesc signals include timing patterns 20 relative to the satellite,, onboard precision clock (which is kept synchronized by a ground station) as well as a na0g Ction message giving the precise orbital positions of the satellites. GPS receivers process the radio signals, Computing ranges to the Cil'S
satellites, rind by triangUlatine` the ,3e rani es, the GPS receiv'er determines its position and its internal clock. error. l:3if1hrent levels of ac Ã_i1 ac ies can be ?c Moved depending, on the 25 techniques deployed, For example, prose ,siria carrier phase observations both from a mobile/remote receiver artd from one orr more fixed-position :cii rer ce stations is often referred to as Real_:-1'ime-Kiiienat.ic or R..1 K, and can prod mu sub cenÃimeteraccuracy.
[0004.1 To gaiii a better trtic] erstandil-ic= of Me accui'ac y levels achievable. by using the 30 (liPS system, it is necessary understand the two typos of signals available from the GP`
satellites. The first type of'signal includes both the. Coarse Acquisition ('C/A3 code.
which modulates Me 1_.I radio signal and the precision (P) code, which modulates boll-, the i_:1 and L2 radio si ;nal a- 1 these a .re pseudcoirand-om digital codes that provide a knmxxn pattern that can he compared to the receiver's version of that pattern. By measuring the CA 02]5]]69 2011-10.05 ._f time.-s tilt required to align the pseudorilndoii digital codes, the UPS
receiver is able to c. anal ute W? tttitalnE}iw;EEi?ttS ps4utfc?range to t-110 satellite, Both the C''A and. P codes liav a a ra.:lati ety long of about 300 me tors (I. m iL E E.as surd and 30 meters 'I

microsecond), respective 1 . Consequently, use of the C ."A code and the P
code yield position data only at a relati5 ely coarse level of'reesoluution.

(110051 The second type of ,i sal utilimd for position deters- in t.tion is the Carrier signal. The term "c:nrriu". as used herein, refers to the dominant spectral component ~,vvhicu remain s it) the radio signal. after t:hL spectral content caused by the mod rl:ated pseudorandom digital LodL< (GA and 1') is r n-ioved. 11e. f_.1 and L2 carrier signals J-ia vl v aveleni thss oi':.abnut f 0 and 241 :entilaaetc.I s, re spw'E\ lv The CAIN
receiver is able to track these. carrier signals, and to doing.. so, n ake measurement,,, of the carrier phase to a snial I fraction Oita complete wavelength, permitiil'3si range measurement to an accuracy of less than a centimeter:.
100061 Satelliue based augmentation systems (S13ASs), such as the F=AA's Wide.
Area Augmentation System (W AS) and tla t. EIEcxf ;<t:Ia iieostut.io a ary Navigation Overlay Service (f_r~.. f (3 9}, broadcast EoI rect:loPa components for r l ?b{al navigation szatellite system (C'sNS', in s:. kid tat the Global Positioning Systc:ria (C,P}')) positioning that Ifl..I ck ionosphere correction maps. fast clock corrector , slow clock correctors and orbit correctors. A single-frequency (511S receiver receives these components over an SBAS
broadcast signal, and using a troposphere model. diffie entiaaily corrects its measured pseudo r ages ultimately improving positioning ac urac .

I0007 One oFthe problems vv oh s xiginw " BA systc.laas is that they are design LCl for use with jingleõtrequenc:y receivers, with which the ability of thee SBAS
system to coIiect 1ollosph1L . `n-ors is one t t dii man lnrni:iatious to achieving hig7heraccualcy, SBA' systern tricdc:l the iulaosphere- as is train shAl and Iii ionosphere delay readings, obtained with a net work oNual-frequency GPS receivers to this shell. SBAS
satellites Own broadcast a vertical delay map of the shell to `313AS-enabled tiPS
receivers so that the single,=frecltle_Etcy GPS receiver can correct ionosphere enrols in its measured pseudo CA 02]5]]69 2011-10.05 ranacs. In die receiver vertical delays are interpolated from the map and scaled acccorcling to an obliquity. lhctor, l lfr..eve the SBAS approach to correcting ionosphere errors is as first order approach and caca have errors exceeding one half meter during normal operation and even tells of meter: during high solar activity. This is particularly two as ionosphere gradients become lartge and the assumption of as thin shelf breaks slow rt.

1()()()8) \\ hitdficaal U,"<S~ Patent No, b . Q7. Ã l:/ ;_liicloses reaktime, single-receiver refa ti a (ills positioning using a technique 4w-,here diftere.ntiaal Correction terms are 14 computed at particular locations and instants of tides, adtuste..d for atmospheric delays and then applied at later instants of time. The later GPS-detined positions are thus deter initied accurately relative to the earlier positions because the ionosphere errors are canceled. out. This patent is at si aned to a common assignee llerewith and is incorporated herein by re bre_nc;. Such relative positioning" aaccurac', is often sufficient for applications not .requiaiing absolute pos Lion#3"i+? accuracy, Such as agricultural and machine control operations Nw,here the primary concern is positioning the equipment relive co its initial location or starting ?oint. For example. agricultural equipment is oaten guided over fields in parallel swaths which arc preferably located .accurately.
relative to each otter, but need not be precisel positioned in a absolute:
GAPS or other earth-based coordinates.

1()1)(91 SB \.S systems are designed to correct only f, l ((./A) pseudo ranges including tllc:; problematic: ionosphere delay component, ha a dual f"requeric receiver can circumvent this need "ices ionosphere c:onvedons by using ! 2(. ) in combination Li'itll either L:l (P) or L l (C/A) 'A) to form the iono. phere-free raatngxes. A
bias, known as the inter-frequency bias. exists between f,lt'll and l 2(P). This bias is diffi_rent for each GPS
space vehicle and takes on a value ranging from a fraction of a meter to a couple of meters. The GPS satellites broadcast this As in a term known rs l`4rtr-Circatrls~f fat but due to word size. This broadcast has a limited resolution of 0.l4meters.
Ri CA 02]5]]69 2011-10.05 100101 There is another bias, On the order of afe de :imeter's, bet meen 1_.I(C.IA) and 1:1(P) that is also satellite dependant, but is not broadcast over the GIP" na igttttion messaan : (today). This As ih the hater-signal group, delay code bias, refer ed to in the r odern z.cd Cfl"S 1Ã.'Ds as \'ariou orguniza-tioris, pailietuiarly the Center for Orbit I: 3elern-#iflatiotn in Europe (CODE) pit-.di:etand maintain St.imat:ea Of 1S r. IL. a,ik a global network of monitoring) l a:raal-l tecitac rkc < GPS re elvers.

100111 r. coed LhSL:ussican of these ht-a,cs can hefound in")u:tl-Fr cir_wnc UPS
Precise Point Positioning with WA.13t:IP'S Cokrecmrs" by I lyunho Rho and l6chard .;arkgel , Navigation Journal of The Institute of Navigation, Summer 2; :07, No. 2, Vol.
54, pp. 139- 15 1.

SUffiii f \Is.Y OF THE INVENTION

is jol .] Although SBAS systems are designed .fm- operation with siÃa gle-fireeluencz' receivers, dual ffequency^ receivers can apply the corrections as well if proper care is taken, "I 'lie advantage of a dual-frequency receiver is that ionosphere-free code and circumvent the need for carrier combinations can be formed that allow the receiver to dic S1BAS ionosphere map. Thus, only SHAS clock and orbit corrector-, are rOCJUired.
10Ã9131 The L I (t. A)/I_:2(f) ionosphere free combination is preferred over 1_,1(Pi 1.2(P) due to the robustrie. hs of tracking I..I(CM over trading of LI(P), The aforementioned inter--h-ccluerac-y and inter--sigi al biases must he taken into account to proper'l = t:ppl ~ ` BAS correction to t o ionosphere-free, combination invol~
ira=g Li(C'/' .) and 1:2(P). The To[) clock correction is not used when uiiliaing ionosphere--free:
observations, however, SB S sisteem assume that is applied so errors in broadcast z 6v will be implicitly present in the S-_BAS corrections.

100141 One additional bias, that must be considered. and % hich is not mentioned to other litenee, tikes into account any k' rlr~#lr:?krl' `Erors:
particularly biases of the S1--BAS system relative to the user's receiver. Such biases have been observed and CA 02]5]]69 2011-10.05 may arise dire to di ererices in the user receiver s tr'LLIcI rnu loops as compared to tracking loops deployed within the receivers conlpns ing, the S13AS net.,wor k. 'I he biases may also co atctin a po ion resulting from systematic, errors in the SBAS
processing algorithms or receive "s. Disclosed here is a method to compute a bias that incltrile as well as biases caused by tin-a iodele; t~L lÃtt inter-siun`ii m ÃI1t .f' IrLC tit It l ~It'fdi 's This method ins o1ves post-processing, data that is col lectCU using a dU it tiequsfl( j' C_3PS receiver < i:t.ir 3tfi; l in :t kliut4'ti location.
W\'L note that dM:t'i:Flt SBAS S\~:tetns (WA AS and E NOS for example i may lead to dil`lerent biases,.

I0 l00-151 Z if bias t rn:I:s must be taken into accot.Ãrst to a:ts liÃeve the highest-level of accuracy in a dtÃal-hccltren s GP Receiver tna:rfal@ op 5R <' S corrections, Only the bits x(i) is readily available to a conventional t_rl'E receiver since it is broadcast in the OPS
navigation r aessa<rc, albeit not at a high resolution (only 0. 14 meters resolution). The remaining biases are not readily available to a conventional GVS receiver. The system 15 of the present invention prof idzs a means to upload and store the needed biases in a table within the recei er. The Upload means can occur on a periodic basis to account for sloe drifts Ãn the biases, or to account ibr bias change (lite to dale:Ilite ti'.pla tl ent ., Also disclosed is a-means to compute the, needed biases.

20 100161 To further improve performance, an additional step can be taken, Rather than relying; sold on a troposphere mode I (arch as the one provided in 1 '1 C
?l. O-22c31)), the approach disclosed here is to estimate actual troposphere perturbati oils to .I
nominal model in real-time as the stag; of it I aini in llte , Such approaches have been taken in prior an, but not in co ibination with a dual-fiequ nc y receiver that ttrn3s 2 ionosphere-five observations that are corrected 5 ~ith S1 AS orbit and clockcor1;eetors, VIII
in a real-time application giving superior li 12~iga6 oia a<IE t ur`acy.

BRIEF DESCRIPTION OFT Hf Dl A% INGS
100171 Fig. I. is a flowchart showing the process ofcalibratir:ca, biases.
10018] Fig. 2 is a flowchart showine the use of the l3- m biases in a (ASS
receiver.

CA 02]5]]69 2011-10.05 DETAILED DESC>'RJP ION Ol TH PREFERRED [`A-I1101.IM11_:N I''S

1. Introduction and f nviron me_ 1t.

Ãlil J As required, detailed embodiments of the present invention are disclosed there to however", it is to be understood that the disclosed embodiments are merely exemplary of the invention, which mas be embodied in various f ar"rras. l'la;
rfi to#
sped tic structural and ttinctional detail: disclosed herein, are not to be interpreted as limiting, but racer ely as a basis à ar the claims and as a representative basis for teaching 11) one skilled in the an to Wously employ the present invention in virtually any.
appropriately detailed structure.

tOO201 Certain terminology will be used in the iollts~vinif description 1-br convenience in reWence only and nd S. iII not b4.. limiting" For example:., up. do~vrs, fi"tarrt, 1 back: right and [Lift refer" to the in ention a 0ri,;.nttd in be ti:l being refcrr"ed W. The words An14'ardl '" and "ulrtVViardk" re.l r to directions toward and 4TG4t1',' horn,rc.,pi;aN .I',, the geometric .t:nterof to embodinMIL being pains th.iL.UI Global navia4rtioa satellite. s stcn s ((;NS S) arc b oddl defined to inelti l4:
(.,PS Galileo (proposed). C:-1.ONAS", (Russia), B idt:a>_r (China). Compass 20 (p10005 . d), IRl' NS (India, proposed). Q?" ' (Japan, proposed) and other current. and future positioning technology using signals from satellites, 4w ith (n.
Without augmentation fro in terrestrial sources.

II. Preferred. Embodiments 100211 'l'he. CAIN contic_al segnlciat Con?l:atrtC' s akllite el;cri l~ biases relative to the inoo phere-free pseudo ranges, P r:p, ., ._= Let C,' P! and .1-12 denote the pseudo ranges observed by a GOS receiver on Ll(t 1'i), Liand 1:, (1') respectively.
Then, the ionosphere tree pseudo rang.- is given by CA 02]5]]69 2011-10-05 1) K'2 1~

i(fl/ f'~ I 57 s A1. `sII-1f, f' 122760 MHz Note that K; - K? 1.

[-0012. Due to group delays and inter-signal delays, there are slight biases between the three obs arz ables Q, A and 112 Ac =ortliugl~, definc the inter-signal code bias, t;rril the ranter-faecfticncy bias rzs, ,,p, as )'I'2_c'3 - p'''p, Bp f r is related to the group del t n known as in the (iPS interface Control Document, f D GP 2 0, by the fihllowing re:(ationsfhip:

B
[00231 Sinec we are concerned with satellite: biases. we shall write the observation equations with the- assumption that the CPS receiver clod.: is lac rfect, and that there are no a mosphere effccts. multipath noise, thermal noise or other error sources aside from satellite clock errors. This greatly sitrtl lil. sthe equations, allowing us to cus on the relevant clock terms. With the assumption of perks t user clock. we eventually solve for this common-mod-c term ~ Shen computing the UPS soltatioar.
Ignoring all brit satellite clock biases. :,e three observation equations become:
C; = I t ' ( 7 - , :
1j C'i t tt~' = R _,- (T,, pp b ., t 2 ;'here B' is geometric ranee (true range. and Tj; is the satellite clock bias relative to ions sphere-f free observation, The U-PS Control Segment uses the CA 02]5]]69 2011-10-05 ionosphere-Free observation When Coil) 1)Utirig its estirnaw of is why c lative to Pie#:,, five, 100241 There are actually two ionosphere- fee combinations of interest, one _? involves the use of a Pt , P_=code combination i hile the other uses the more reliable 1)_i combination. I hese are:

R
c71~~
li`rt., K,,.1) == R - T-Kii Cdr, _.;

n/a n-v we have used the relation slitfl " K2 B to s mplii4'.

100251 Both ionosphere-l eee observations, are, as their n me irriplies, tree of any ionosphere induced delays. They are desirable :since My enable a receiver to overcome.
one of the largest remaining sources of errcar in single frequency :'BAS
positioning: the 1 un-modeled ionsspherc. SBAS cannot provide a perfect Knosphere map, and it is better to siirtrlals cancel out the ionosphere efIc s using these ionosphere-free observations.
10Ã9261 When SBAS sends cloc: correctors, the clock correctors are relative to the C:`,obiervable, not the ionosphere-We observations. In computing these corrections, 49 the SlIAS s =st: n assumes (see R i t .''?/D0 ``2UD) that dits user receiver will ripply the GPS broadcast clock model. as well as the broadcast arotr l-delay. The 1 now GH
supersca ipt''i=ri's" indicates that and are estimates provided by the GPS
C ontrol St n-ment rattler than the true viiities. The GPS a avigatioran-tessane. which is modulated on the CPS transmitted sit>nals, contains coefficients of a time-bused 25 (?t31\'noi211 llttltl The navigation message contains a single word providing the CA 02]5]]69 2011-10-05 group-delay.. Each satellite sends its respective, _saÃe[] ite-spec:ilc:
values for the aabov : quantities.

[OO27] The SI3AS clock cor'rectioÃa although provided by SBAS systems as a single quantity, can be broken alaawn into its individual constituents as follows:

1OO281 In the above equation, it is assum e.d that S13AS does a pertc:ct lob of correcting the clock errors in the (71 observable. In other words. the SBAS
system ma[ e.s no errors when pre ic.tÃr1u 714 .CS i and = (which. 1`,='.L note, it does as a lump d clocked correction rather than separately as we have showvn), While this i , not precisely true, it i a sufficient =sunrtption for the. purpose ofex.planation, and as will later be shown, long-term systematic errors are treated as biases that we can compute and eliminate.

[()()291 According tothe SBAS specific tion, the SEAS clock Correction is added to the C, observable while the conventions outlined in the. 1C.13-CUPS-200 are Rd Wed e=v ith r .spcc::t to use jai aait_I the ""'BAS-corrected pseado range becomes 5.1f15eo re:c er Li ( t) t seitdo lunge i i =1 'SR.r`
an substituting in the constituents of CO and f; j,,,S we get 5114 col rcc/ed L /(C.' 1) tr7:~i-'t{ti{ ! an e ::.
I . ~,,;: r ISt f i R - (' -- 3 . C r s IPS T.
:_5 =1 As can be seen. if SEAS does a perfect job correcting the clocks, the corrected pseudo range becomes the true geometric range / (ignoi ingg all non-clock errors) as desired.

CA 02]5]]69 2011-10-05 - l tl --100301 Now, consider a similar application of the. SFIAS correctors to the r sa r based on rtr me] Here, tc::cording to the ie; nosph re--iee.e combination IC-D-GI S-200. w =e Should not subtract at least that is the case when a ink Me ionosphere-tree obsert shale. will denote the SEAS-cor'r'ected ir_anospheae tree observable as ll ) 3t i -V Al", { ..; flr J'-Y =A
,.s.'g. 3 I ,. F + j-C=. J>>iS S .c 3;~

100311 The terÃta B s > >~rr , is a. has thtat must lie added for proper application of SBA ; writ Curs to the t`-.` / P--ba d ionosphere tt.`e' :onihinwion. Its neLd (und l 4 l0 value) will becoia-te evident Ãn the subsequent dt rivtaÃion. Expand ng terÃ1as we `et ~.urr. = rre.f T 7 t R ~T,f- kt` a l .9 ) + 1'a ~{ Ã t? i3 t .~~`= / -'' j-};~ if R (K]

In order to get the c; esired result.. nitisi. he taken as thllows:
' \,> ith this, we have l 1' 1 We see tliat Ai; (c;,r) account for the intef s.t?ta rl e otle. bi } . ZKf :c;j, as weI I ias any errorss 0r' round-ott in tile. precisi :n of 1ll0321 One :more: bits that. timust be included is one nt t included in previous 20 literature. vkili.ich we shall denote: as >".,=Frz,, .= hats bi }s accounts for dilibreiac-es between the code tr'aacking-loops Ãn tl7e SHAS's networ.-k of (iPS receivers and code trtckiirat? loops iri t user`s GPS receiver (diifet'e'nt r- uitipath naitigtat on teduii ues and receiver ni'o ip-delavs can result. in dili-t're-nt biases). It also aaccouitits lo'r an other-.

irra,I :rl rrt.ititin s e -.it c biases betvv'een Mae SEAS sti<stem and user receivrer. One v- ay 2? t o descriti( f3(...r 1, ;s. Ãs } bits thr t models the difi r:nce b ;tween the pseudo rang :

CA 02]5]]69 2011-10.05 -ll-=
measured in a receiver that would be corrected perfectly, bySBAS, and the pseudo range.
lmmeasured in a receiver that is actually.-- being daplo ed.. 'I 'lie total bias that ft-just be taken into account when using SEAS corrector with t:he. ionosphere-lies: `j / 112 combination is tt 100331 In the exemplary embodiment, this bias, is stored in the non-vokltile memory of the user's CiPS receiver, one for each OPS space vehicle.
The biases have been observed to be nead' . onSttat, varying slws:4 I y over time and only- takilizg di'anuatiL steps upon replacernem. of a spate v hic.IL. E:;,arlsec]ncntl\a, frequent Liploads of biases to receiver non-volati1e memory is not necessary. 11ploaci.s c an # ~
done t t to periodicbads which will typically he on the order Of several months to a year, ['lases should be monitored for substantial changes so that new.k bias tables can be made 4lv-a.ilable ; e. note tii.at of t; c.olfiponelit oI itto hiases, tilt bits.
is monitored In the Center for Orbit l)etertnination in Europe (CODE) %,,'here results are made public via the Internet. However, Or best results it is recommended that biases are.
monitored and computed using the exact tyype. of GPS receivers for which the bias will be utilized.
Particularly since CODE, uses a network of receivers from different r annufacturers to compute their bias estimates based on an average oi=all contributing receiver data.

[0034[ Rather Man compute f1?0; :i, as the sum of individual components, it Cm be computed as one lumped bias. This is accomplished by replacing Bin: r_:r>
with B:t)l-i in the equation for The process ofcomputing2 will be referred to as calibration mode.

100351 During, calibration mode, at receiver is situated in a known location and configured to gather c:arr iwi-sll\~lcitl~ed, ionosphere-free pseudo ranges correct d with SEAS orbit a nd clock correctors. Smoothing can tak . place indefinitely since there is no divergence between code and carrier evith the iOilosphere free combinations.

p00361 The carrier--smoothed, ionosp#lca;e 11ce pseudo ranges, corrected with C1.3AS are computed as CA 02]5]]69 2011-10-05 l2tKC
where ( __ F~ t rr ;
fLf1d), l i.~:!<1err the is the is~1rrier-=53t oot~lle , it yE:ao;,:3her ei'i:.C; pseudo range j . ' is the GI'S broadcast. S V clock model is the user clock estimate, common to all observation',,, Am is the SBAS dock correction (both fast and slow) rr z Illtatr't l is a nominal model of the troposphere delctt_ as tQt e_xttm le, the model in RTCA/DO 229, Appendix A

10037. A s residual is formed 6or each satellit . by subtrtctine the ue:otnetri (truth) r(Ittge of the ratc:lhte Rom f ' ! F he truth rtrtga LerÃ;, is computed using the O PS broadcast ephemeris (correcting. with SHAS orbit correctors) and the known receiver location, ideal ly- adjusted the solid Earth tide variations in tltitude.

[00381 The residual includes the desired Was, 8plus other errorterms. That is residual P" A Other Other is all other sources of error aside from those constant t'a'or making tip The term -OtheV" wVOL.Jld include.. the following:

2t Ã_=t' -modeled troposphere -'I'll :s is the troposphere perturbations from the Ãnodel, We- attempt to e timate this source of error in post processing.

Errors in the user clock. ` : e l he e. Ã oÃrs are common to all satellites and are removed in post processing by estimating it and di ferc.:ttting it aw-vay.

Me varying errors in 7W,-- T:hese are dtr to the tact. the SBA system is not .2,`? 114 t ]C:s':t.. MY '.Y are treated as noise. Note that any constant errors that persist f r days will be. Itimpcd inttoB-- -';:=
l:.ilects of Solid Earth tides. These are actually not errors but variations in the receiver location (up to 30 cm) due to solid Lartlt tide alt.Ã!_:t.i3 Ãtions in receiver CA 02]5]]69 2011-10.05 altitude, A triode-l of solid f_:arth tides should be utilized to adjust the receiver truth location accordingly, V a model is not used, data should be taken over ruultil ie days so that the [Earth tidal effects can he aver aged out.

Muitipath and other noise --- the antenna. should be as choke ring or equivalent and placed in 4 good location.

((mm] Residual terms are gathered, ideall~ over multiple days R--or beticr trd . i iging away of nd c s:ffi"c>ts, Post processing or the residuals then yiel cis the desired biases. in the crudest furrn of post-processing, residual data is simply p'tottcd and tss s r:anatic shifts in residuals mt: visually determined. 1 hc= eye is c;
.raer dl' good at averaging. out time-varying effects, such al, those mentioned above.

1O04D:1 A more systematic: approach to calibrating, the biases is to use a .K:alnian filter or a feast-squares method to estimate and reprove the effects of un-modeled troposphere errors, common clock errors, and. to efficefively time-av e1a; e the data. The state vector of a Kalman .filter is comprised of a troposphere perturbation state .and a clock state. both of which are allowed to vary (a random-walk process}, and the y Aj, biases, one per sdat.ellitc, each of which is assumed to he constant in the Kalman; grocers model, The troposphere state: is a perturbation from the model of the trtposphera and 2C) its of feet on the residual observations is through a mapping function based on satellite e1ev ation anf e. By estianttting and rcrnavina twpo he and clod, effects, the Whom him filter is able:. to effectively e timate the bias terra (essentially these biases are a time-average of what is left in the.residual errors after removing troposphere and clod.
(I ffsets).

1.010411 As mentioned encompasses inaccuracies in the assumed I~( as well as 'HAS sysie n-to-tsar rc.c ivcr hr asc.& Clearly, portions ofl3;;r.:
such as the 1St r.;, . or GD contriibutton) Can he dcteriaiaraed in advance by other meth ads, ffo example 1St.',r=, =i can be ctete_nnhied s ra-al 4s, by averaging PI-Q. In thi's casel the known bias con r'ibutions are removed up--front during the calibration process.
Only the remaining portions of then calctalated. Gad, fi~lfr,ia in c=alibratio.ra:.

these remaining portions art-, summed together ?h ith the 1~fli?1L'tt biases to yield the desired total bias. 100421 Having performed the bits c ;librations, a table of satellite biases (t.

separate 13x0 a-.i. for each satellite) is made available for upload to receivers in the Ãield.
This could be via the Internet (. itfh aj web based) server for example) or the bias table could be loaded onto a memory stick that is plug; ed into the individual receivers. The user may manually enter the bias table into 1.110 receiver d 1R-ML is ding a ecammand Over, one of the receaa eÃ-'s RS 232 cnm ports, or the process may be automated, such as simply comiecting the receiver to a PC and letting soft are take oVei-, The receiver then stores the biases in non-volatile memory for subsequent use. Other IpprOadle,;
U111 be taken which are obvious to those skilled in the tart.. Biases are re-computed aai-id made available On a periodic basis Or as it satellites cars' launcheL.

100431 Fig, 1 shoe s the process of calibrating biases. 13lock_s 101 through provide a laugh-level summary ot`the process steps described above. ;light variations to these steps and their sequence will he apparent to [hose skilled in the art and. within the scope of the present invention, V step 101 a re er-aiac-e receiver is situated at a known location. Ionosphere--free, carrier-smoothed pseudo ranges are computed at 102' and S BAS corrections are app icd at 10 . Residual errors la tt~een true ranges and `3BA S-col i ected ranges are computed and output at 104 and the biases are determined at 105.

1Ã 0441 In summary, a single i e-c :iv er situ .ted at a known location can be used to calibrate biases (producing .a table: of 13-i, i -ac at 106) that can later be provided (by some rci~:atil. } to large number of receivers in the f eld. he process of .Firs. 2 is done periodically to monitor and capture slow dean es in biases:. O is done whensatellites enter the constellation.

0 10 0 451 With the biases f nowii and accounted R )r.,. the SBA Orbit and clock are applied to the ionosphere-t.re range observations r suiting in. improved Mlracy in a dual-frequency ()PS receiver. As was the cane during calibration, Smoothing of CA 02]5]]69 2011-10.05 -- l . f ionosphere-free code against iot crsphere--Ree e.ir i r t tbr example, a f-1 atchà fiRerj is 4,-mpto cd leer best results, ;nii;:thin4e in take f~l<ai:.ta iÃlilctira t:It since there is no divergence betty en code and carrier with the iortosl aa.rc ti we.
combinations. his is Uttl iko aÃ. single-frequency receiver htv smoothsÃa inierz <t:ls must be limited as à esu It, of codc/c:art tt= citz: rgc.a t e. In practice, as two hoar smoothing \ i in ioty would fypic;:ffly be suf'ficient 1-or m in'E aalzplica.ions.

114$461 Fig. '1: outlines the use of hz B- y;1i ts. biases in a GNUS receiver.
As shoat n in block 121,. the biases are first uplciaded into the receiver, During operation the GNSS
I0 rt.ccio :r competes ionospli ao i;e~. pseudo ranges at 122 and fl-ion (optionafly) applies carrier smoothing to these ranges at 123. Corrections are then made using,, SBA S clock correctors at 124, the biases at 125 and modeled or- estimated troposphere delays at step 12& Using the corrected pscudo :rate,, c_y, :a location is then computed as shown in 127. S13 AS ephemeris corrections are applied as %vell. In fact, all SBAS
correction s can l5 be applied as described in IMAM() t_ A.SDO 229D. Appendix A with the of the ionosphere corrections. The over madorr difference is tha.at We haVC
ili`Codticc-d the l rrai, Ã. bias corrections and we no longer apply 'ton wince it is not Used with ionosphere.-free pseudo ranges.

20 100471 in the exemplary embodiment of the present invention an aaddidonai step is taken to improve accuracy; This step is the estimation o resà dal troposphere content in trae observed pseudo ranges. Haav m = removed ioransphei ci-ror using ionosphere- free combinations and correcting clock and orbit errors aisinu S13AS corrections, troposphere model en ors start to become relevant and limit overall accuracy. However, with i sufficient numbers of satellites (5 or snore), perturbations in the troposphere relative to the model can be estiniaatt d in real =titne.

10045 A state that m odc-Is a troposphere zenith delay perturb,, doÃa is estimated b a aK.alÃtaan filter that also simultaneously estimates receiver lo<ati(m, clock, and other 30, navigation quantities of interm in another embodiment, the Kalman titter can igno re a nominal mod.e.l alts gether and estimate the entire zenith troposphere, not just a CA 02]5]]69 2011-10.05 per(urbation to a nominal mod&. In constructing the aln ia:n Ilhtek', measurements are assumed to be impacted by the troposphere zenith delay staled by a itiapptn`r function that maps enith into a slant delay based on satellite elevation angle.

100491 A Kalman Inter state vector will consist of'position (v n '), time, /.
and tropo phere. r pWs possibly other quantities such as velocity or past positions.
=

1 t 100501 The Kalman residrial. fe==~ir srteilitte.,j. i deiir,e i as resiACI R, i tS ¾ r .;::tr:i' ;3'tri ere the-<1ih satellite's measured pseudo range 15 and R;'."-" they ~l katellite: s prodickd range, whore the prediction is based on the current state. of the K hntan inter.

[005II The Kal rniain inter is based on a I ine arization of the on-l inear range 20 equations. The partials of he residual th respect nr=1t~~=t=

.}}\\\\\\\\\\\\\\\\iw.\\\õ.........................
........................... ................

CA 02]5]]69 2011-10-05 1I ` = ;s tf r.~ t'.. Ali, = is am? 3 row ot'\\-it,:t is Irequently ref tired to as the design matrix and is the matrix needed for Mum filter dl sign, or feast squaae s solution. The tar;st 3 elements of Ire; rr;; , at' ,~
are aa. unit vector from sa:z 1litL j to the current: location X. and ,`s..:/' is a troposphere à wpping 11inction for satellite,I. 1"he clock, i; is mapped through unit.: (the 4;;s element of W ).

(Ã 71 r troposph r~ .a a,.ap iÃat Ã'r!rac:tÃon, is taken as ' Al where is the elevation angle of sate] lite , .

141551 The Kalman time-tapclate of troposphere state can fallow a simple linear model which exponentially correlates over time. The following equation shows the Kalman process mod d for the update of the tri pospher state r , We time k. to time R L exp 1>

HHlere, nr(k) is tin it We arise , is the I .,aim to sample period, Te is the troposphere tiira~ ct~Ã'r latitara poriod, and P is the tropci,iphCre Mato-Lorrz:lation power. Typical ; .altars, for Te aare one to Several hours and P} is on the order o t t ntim tsr. since r(l) represents a perturbation from nominal tropo ~lphere.

[00541 he equations given above are stattieient, for one skilled in die- an, to design a Ka nian Sher estimator Or to navigation state that includes ia,tropospher:
My estimate in addition to receiver location and receiver clock.

10Ã1551 It is to he and erstood that the invention can be embodied in various #orn"as, and is not to he limited to the examples discussed above.

Claims (18)

1. A method of determining the location of a global navigation satellite system (GNSS) receiver including a receiver memory, which method includes the steps of:
computing ionosphere-free pseudo ranges using code and carrier phases of GNSS
signals;
correcting said ionosphere-free pseudo ranges using satellite-based augmentation system (SBAS) clock and orbit corrections;
providing a microprocessor controller connected to said GNSS receiver;
computing a table of biases to further correct said pseudo ranges;
uploading into said receiver memory biases; and computing locations of said receiver using said corrected pseudo ranges and said biases.

2. The method of claim 1, which includes the additional step of:
carrier-smoothing said pseudo ranges at a smoothing interval of greater than one half hour.

3. The method of claim 1 wherein the troposphere is estimated as the state of a Kalman filter.

4. The method of claim 1 wherein said GNSS receiver comprises a dual frequency receiver using one of:
L2(P) and L1(P); and L2(P) and L1(C/A).

5. The method of claim 4 wherein the SBAS-corrected L1(C/A) pseudo range =

= R

wherein:
the GPS broadcast clock model = ;
the broadcast group-delay = ;
the superscript "GPS" indicates that and are estimates provided by the GPS Control Segment rather than the true values;
the GPS navigation message, which is modulated on the GPS transmitted signals, contains coefficients of a time-based polynomial fitting ;

the navigation message contains a single word providing the group-delay, ;
each satellite sends its respective, satellite-specific values for the above quantities; and the SBAS clock correction, T SBAS, can be broken down into its individual constituents as follows: .

6. The method of claim 5 wherein:

.

7. The method of claim 6 wherein:

; and .

8. The method of claim 1, which includes the additional step of:
calibrating the biases using a Kalman filter or a least squares method to:
estimate and remove the effects of un-modeled troposphere errors and clock errors;
and to effectively time-average the data.

9. The method of claim 1, which includes the additional step of:
providing a separate B TOTAL value for each of multiple GNSS satellites.

10. A method of determining the location of a global navigation satellite system (GNSS) receiver, which method includes the steps of:
computing ionosphere-free pseudo ranges using code and carrier phases of GNSS
signals;
correcting said ionosphere-free pseudo ranges using satellite-based augmentation system (SBAS) clock and orbit corrections;
providing a microprocessor controller connected to said GNSS receiver;
storing in said microprocessor controller a table of pseudo range correction biases;
correcting said computed pseudo ranges using said biases;
carrier-smoothing said pseudo ranges at a smoothing interval of greater than one half hour;
estimating the troposphere as the state of a Kalman filter;
using either: L2(P) and L1(P); or L2(P) and L1(C/A);
said GNSS receiver comprising a dual frequency receiver;
correcting the L1(C/A) pseudo range according to the formula = R
wherein: the GPS broadcast clock model = ;
the broadcast group-delay = ;
the superscript "GPS" indicates that and are estimates provided by the GPS Control Segment rather than the true values;
the GPS navigation message, which is modulated on the GPS
transmitted signals, contains coefficients of a time-based polynomial fitting ;

the navigation message contains a single word providing the group-delay, ;
each satellite sends its respective, satellite-specific values for the above quantities; and the SBAS clock correction, T SBAS, can be broken down into its individual constituents as follows:

calibrating the biases using a Kalman filter or a least squares method to:
estimate and remove the effects of un-modeled troposphere errors and clock errors;
and to effectively time-average the data; and providing a separate B TOTAL value for each of multiple GNSS satellites.

11. A GNSS receiver system for receiving signals from GNSS
satellites, including at least one SBAS satellite, which receiver system includes:
a receiver including a nonvolatile memory adapted for storing a table of bias corrections for at least one GNSS satellite;
each bias correction including at least one of a group delay correction component and an inter-signal bias correction component; and means to compute an ionosphere-free pseudo range using data broadcast in the SBAS signal as well as data in the bias correction table.

12. The GNSS receiver system of claim 11, which includes:
a Kalman filter adapted for estimating the troposphere as its state.

13. The system of claim 11, which includes:
said GNSS receiver comprising a dual frequency receiver using one of:
L2(P) and L1(P); and L2(P) and L1(C/A).

14. The system of claim 11 wherein the L1(C/A) pseudo range is SBAS-corrected according to the formula:

= R
wherein: the GPS broadcast clock model = ;
the broadcast group-delay = ;
the superscript "GPS" indicates that and are estimates provided by the GPS Control Segment rather than the true values;
the GPS navigation message, which is modulated on the GPS
transmitted signals, contains coefficients of a time-based polynomial fitting ;
the navigation message contains a single word providing the group-delay, ;
each satellite sends its respective, satellite-specific values for the above quantities; and the SBAS clock correction, T SBAS, can be broken down into its individual constituents as follows:

15. The system of claim 11 , which includes means for calibrating the biases using either:

a Kalman filter or a least squares method to: estimate and remove the effects of un-modeled troposphere errors and clock errors; and to effectively time-average the data.

16. The method of claim 11, which includes means for providing a separate B TOTAL value for each of multiple GNSS satellites.

17. A storage medium encoded with a machine-readable computer program code, the code including instructions for determining the location of a GNSS
receiver, which method comprises the steps of:
computing ionosphere-free pseudo ranges using code and carrier phases of GNSS
signals;
correcting said ionosphere-free pseudo ranges using satellite-based augmentation system (SBAS) clock and orbit corrections;
providing a microprocessor controller connected to said GNSS receiver;
storing in said microprocessor controller a table of pseudo range correction biases;
and correcting said computed pseudo ranges using said biases.

18. A computer data signal comprising code configured to cause a processor to implement a method for determining the location of a GNSS
receiver, which method comprises the steps of:
computing ionosphere-free pseudo ranges using code and carrier phases of GNSS
signals;
correcting said ionosphere-free pseudo ranges using satellite-based augmentation system (SBAS) clock and orbit corrections;
providing a microprocessor controller connected to said GNSS receiver;
storing in said microprocessor controller a table of pseudo range correction biases;
and correcting said computed pseudo ranges using said biases.