MAI-Mitigation and Near-Far-Resistance Architectures for GNSS Receivers

Transcription

1 Journal of Global Postonng Systems (23) Vol.2, No. : MAI-Mtgaton and Near-Far-Resstance Archtectures for GNSS Recevers Z. Fu () and J. Wang (2) ()Insttute of Communcaton and Navgaton, German Aerospace Center DLR, Muenchner Str. 2, D Wesslng, Germany. Current contact: E-mal: zongzhou_fu@yahoo.com (2)School of Surveyng & Spatal Informaton Systems, he Unversty of New South Wales, Sydney, NSW 252, Australa. Receved: 3 October 22 / Accepted: 4 February 23 Abstract. Multpath, MAI (Multple Access Interference) and near-far effects are the three man nfluences on the performance of CDMA-based communcaton and navgaton systems. A great deal of research has been conducted to develop advanced sgnal processng algorthms and novel recever structures useful for mtgaton of these effects n moble land wreless communcaton systems, such as UMS. Although the multpath effects on PRN code rangng n GNSS have been nvestgated for about two decades, the MAI and near-far effects have only been dscussed n pseudolte applcatons. In ths paper, the mparments of the satellte-moble recever channel wth multpath-selectve fadng, and shadowng/attenuaton effects by objects such as trees/forests and buldngs, are theoretcally analysed, under a more general and practcal defnton of the near-far effect. he MAI-mtgaton and near-far resstant recever structures for Galleo/GNSS applcatons are presented. he prncples of such recever structures and ther applcatons n GNSS are dscussed. Both theoretcal analyses and computer smulatons are presented and show the applcablty of the proposed recever structures. ey words: Multpath, Mtgaton, Sequental/Parallel Interference Cancellaton. Introducton Snce the nventon of the Drect Sequence Spread Spectrum (DS/SS) technology, the traonal correlaton/matched flter recever structure has been used for the detecton/tracng of DS/SS sgnals. Despte extensve developments n DS-CDMA (Drect Sequence Code Dvson Multple Access)-based communcatons (e.g. UMS), the classcal recever structure s stll used n current DS-CDMA-based global navgaton satellte system (GNSS) recevers (e.g., aplan, 996, Parnson & Spler, 996), and seems to be contnuously used n the future Galleo/GNSS recevers. Whle the traonal recever structures perform optmally n adonal whte Gaussan nose (AWGN) envronments (e.g., Proas, 996), when used for GNSS sgnal tracng/detecton for moble land applcatons, MAI and near-far problems may arse. herefore, the conon of the conventonal correlator/matched flter s sub-optmal. MAI refers to the nterference between DS-CDMA satellte sgnals, whch s nherent to CDMA systems. hs nterference s the result of the random tme transmsson delays between sgnals, whch mae t mpossble to desgn the PRN codes assgned to each satellte to be completely orthogonal. Whle the MAI caused by any satellte sgnal s generally small, as the number of satelltes ncreases, the MAI level ncreases, thus degradng the recepton qualty of all the GNSS sgnals. he MAI s conventonally treated as whte Gaussan nose. he near-far problem occurs when the power of the sgnal receved from one transmtter s so strong that the sgnal receved from other transmtter s completely jammed. hs s the case when the sgnals from the dfferent satelltes arrve at the recever wth wdely varyng power level dspartes. It s commonly thought that the dspartes of the power levels of the receved sgnals are caused by the large dfferences n dstance between the transmtter and recever. For GNSS, such as GPS, there s no serous near-far problem snce the satelltes are all

2 28 Journal of Global Postonng Systems roughly at the same range, and the receved sgnal levels are assumed to be nearly equal (e.g., Parnson & Spler, 996). Multpath effects n satellte-based PRN rangng systems have been nvestgated for about two decades, and many methods to mtgate multpath effect exst today (e.g., Van Nee, 997). However, lttle attenton has been pad to MAI and near-far effects, and they are commonly neglected n the desgn of conventonal GNSS recevers. Verdu (986) has touched brefly upon ths subject and concluded that the near-far effect could be a bg problem wth pseudolte applcatons, though ths can be solved by the use of DMA sgnal transmsson, for example. But the near-far effect s not a bg problem for GNSS recevers, because satellte sgnals wth large receved power level dfferences generally also have large Doppler shft dfferences, whch exceed the tracng loop bandwh. However, the near-far ssue needs further nvestgaton. Frst, a GNSS recever always wors n the modes of acquston tracng loss-of-loc reacquston tracng. Whle the near-far effects can be compensated by the great Doppler dfferences n the recever tracng mode, they cannot be avoded or compensated at all n the recever acquston mode. hs s because n general the acquston bandwh of a tracng loop worng n an acquston mode s much larger than any Doppler shft, whch s ntentonally desgned for the purpose of fast acquston. Second, the so-called general case assumed prevously s actually not general, accordng to our analyss n urban and suburban moble GNSS applcaton envronments. he mass maret demands the producton of low-cost recevers wth optmal performance that they are able to operate n envronments where most consumers lve, travel and wor, such as n a movng car, urban, suburban or even ndoor areas. It s therefore necessary to develop new technologes for advanced GNSS recevers, offerng the capabltes for the MAI-mtgaton, near-far and multpath-resstance. It s also mportant to nvestgate new GNSS/Galleo recever structures wth the better performance at a low cost, snce the new satellte navgaton system Galleo and the GPS modernsaton program wll use new sgnal structures or parameters (e.g., EIRP, PRN chp rate etc.) for the satellte rangng sgnals (e.g., Spler, 999, Hen et al., 2). One of the specal attrbutes of the proposed new sgnal structures n the GPS modernsaton and new system Galleo, s the use of plot sgnals. hs provson may allow for the mplementaton of advanced recever structures, wth smple or moderate computatonal complexty. In the followng secton, the mparments of the satelltemoble recever channel wth multpath-selectve fadng, and shadowng/attenuaton effects by trees/forests, buldngs etc., are theoretcally analysed. he problems wth the conventonal correlaton recevers are dscussed n Secton 3. he prncples of the proposed recever structures, whch have capabltes of MAI-mtgaton and near-far resstance, are gven n Secton 4; Fnally, the computer smulaton results, and dscussons on the possble enhanced applcatons are presented. 2 GNSS CDMA-Channel Model GNSS, for example, GPS and Galleo, s an asynchronous CDMA system,.e., the rangng sgnals are randomly delayed from one another because of dfferent transmtter postons and dfferent propagaton paths (channels). he GNSS CDMA channel model s llustrated n Fg.. S atellt Propagaton Channel s (t ) S atellt Propagaton Channel s (t ) Σ r( GNSS Recever s (t ) S atellt Propagaton Channel Fg. GNSS CDMA channel model

3 Z. Fu and J. Wang : MAI-Mtgaton and Near-Far Resstance Archtectures 29 Each satellte sgnal passes through a dfferent path to arrve at a recever. A transmtter onboard each GNSS satellte transmts a bnary baseband DS-CDMA sgnal, derved by multplyng a bnary level (±) nformaton sgnal wth a satellte-specfc bnary level spreadng sequence c (. he spreadng sequence c ( exhbts the constant modulus property: c ( () where denotes the absolute value. he -th transmtted sgnal from -th satellte s ( s gven by: s ( c d ( c ( cos( ω ( t τ ) φ ) (2) where d ( s the nformaton bearng the data sgnal wth the symbol duraton b. recever) s llustrated n Fg. 2 (e.g., aplan, 996). he fgure shows a ban of correlators. Here each PRN code waveform s regenerated locally, and correlated wth the receved, summed, and IF-downconverted sgnals r RF ( n each separated correlator branch. A conventonal recever follows the detecton/tracng strategy, that each branch sees only a certan desred satellte sgnal and processes other satellte sgnal nterference (termed as MAI), as the unstructured channel nose (termed as AWGN). o smplfy the analyss, wthout loss of generalty, t s assumed here that there s no multpath effect and that the data modulaton s Bnary Phase Shft eyng (BPS). he receved and summed sgnals r RF ( s gven by equaton (3). A smple conventonal recever structure, as stated above, currently used for tracng the PRN codes of satellte, satellte 2,..., satellte ( s 2 n a modern GPS s ( s 2 ( s ( r RF ( b o ( ) Z ( cos( ω t ) c φ c ( b o ( ) Z 2 ( Recever Processor cos( ω t 2 ) c 2 φ c 2 ( b o ( ) Z ( cos( ω c t φ ) c ( Fg. 2 A smplfed ban of a conventonal GNSS recever structure for recepton of the sgnals, s (, s 2(,..., s (, from satellte, satellte 2,..., satellte, respectvely

4 3 Journal of Global Postonng Systems rrf ( 2P c ( t τ ) d( t τ )cos( ωc ( t τ ) φ ) nrf ( where P he average receved sgnal power n th satellte, he number of satelltes beng n vew, τ he tme delay of the th satellte through -th propagaton path,2,...,, ω c φ n RF ( he th satellte s carrer frequency, whch s equal to ω c ω c ω c, wth ω c beng the nomnal carrer frequency and ω c the carrer frequency offset, he phase of the th satellte s carrer sgnal, he AWGN resultng from the RF front end. (3) Assumng that the th satellte s the desred one n the th branch of the ban of correlators, and assumng perfect carrer recovery (.e. ω c, φ ), the baseband sgnal r( at the moble unt s gven by: r( 2P c ( t τ ) d( t τ ) 2P cos( ω τ φ ) c ( t τ )) d( t τ ) n( (4) where n( s the equvalent AWGN at baseband. Assumng perfect code synchronzaton of th correlator (τ ), the output of the th correlator s obtaned as: Z b b b b b b r( c ( 2P d[ b 2P cos( ω τ φ ) n( c ( c 2P d MAI n c ( c ( ] b d ( t τ ) c ( t τ ) c ( (5) wth b beng the duraton of the data symbol of the nformaton waveform d(. As stated above, because the satellte-recever s an asynchronous ln, the term wthn the frst par of bracets s unty. As a result, Equaton (5) can be wrtten as: Z MAI n 2P d MAI n b b 2P cos( ω τ φ ) b c b n( c ( d ( t τ ) c ( t τ ) c ( where the frst term n Equaton (6) s the desred sgnal component, the second term s due to MAI, and the thrd term s AWGN. From Equaton (6), the MAI and near-far problems can be easly understood. MAI problem he conventonal correlaton sees only the desred sgnal for detecton and tracng, and the non-zero crosscorrelaton-caused MAI would be zero f the PRN codes are desgned to be perfectly orthogonal for random tme delay τ. Unfortunately, t s hardly possble to acheve ths deal result for the satellte-recever asynchronous lns and therefore MAI always exsts. In practce, PRN codes wth near-deal propertes (good auto- and crosscorrelaton functons) are sought after (e.g., Spler, 999). Whle the MAI caused by any satellte sgnal s generally small, as shown n Equaton (6), an ncrease n the number of satelltes results n MAI level ncreases. hus the degradaton n the recepton qualty of all lns n (6)

5 Z. Fu and J. Wang : MAI-Mtgaton and Near-Far Resstance Archtectures 3 GNSS requres consderaton n the GPS/Galleo Overlay scenaro (e.g., Hen et al., 2). When a large number of MAI sgnals are receved wth almost equal power levels, the MAI appears to be Gaussan accordng to the Central Lmt heorem, and almost whte wthn the band of nterest. hus, the conventonal recevers may approach ther optmal performance. Unfortunately, for GNSS applcatons, only a moderate number of satellte sgnals are receved smultaneously and the sgnal power levels may not be the same. herefore, the conventonal MAI-AWGN assumpton s not realstc and the correlaton recever structure s not optmal. Near-far problem As shown n Fgs. 3, 4 and 5, sgnal propagaton from dfferent satelltes s attenuated to dfferent power levels by a varety of objects, such as trees, forest, and buldngs. he conventonal assumpton that all satellte sgnals have equal power levels s unrealstc n practcal GNSS applcaton envronments. Car/Moble User Fg. 5 Power level dsperson caused by multpath fadng 3 MAI-Mtgaton and Near-Far-Resstance Recever Archtectures 3. Sequental Interference Cancellaton SIC Algorthm he prncple of sequental nterference cancellaton (SIC) s shown n Fg. 6. Car/Moble User θ d Forest or rees Dgtal Sgnal from ADC Receved Dgtal Sgnal Buffer Each sat Sgnal Fnd Strongest sat Sgnal for Cancellaton Fg. 3 Power level dsperson caused by tree attenuaton Regenerate Strongest sat. Sgnal Subtract Renegrated Strongest sat. From Buffer Fg. 4 Power level dsperson caused by dfferent path attenuaton Fg. 6 Bloc Dagram of the Sequental Interference Cancellaton (SIC) All the sgnals are estmated at each teraton of the scheme. he sgnal wth the largest power s then regenerated and subtracted from the buffered receved sgnal. he remanng sgnals are now re-estmated, and the new strongest satellte sgnal s selected, regenerated, and subtracted. he process contnues untl all the sgnals have been recovered or the maxmum number of cancellatons s reached. After satelltes through -

6 32 Journal of Global Postonng Systems have been removed, the decson statstc for the th satellte s: z ( ) b r ( ) ( a ( t τ ) (7) where r () s the receved sgnal after satelltes through - have been cancelled, whch s gven by: ( ) 2 ( ) r ( r( Z a ( t τ )cos( ω θ b j (8) It has been shown that SIC s very robust to dverse power levels (e.g., Patel & Holtman, 994). hs s due to the strongest satellte sgnals all beng cancelled from the receved waveform. he SIC s consdered as one of the smplest forms of MAI-mtgaton and near-far resstance recever structures. However, the algorthm for the cancellaton must perform all the cancellatons whle mantanng the necessary navgaton data rate. Obvously, the larger the number of satellte sgnals, the longer the processng tme. 3.2 Parallel Interference Cancellaton - PIC Algorthm he prncple of the Parallel Interference Cancellaton (PIC) s shown n Fg. 7. In Stage, a ban of correlators correlate all the satellte sgnals receved. hen, each satellte sgnal s estmated and regenerated. In the next stage, a new estmate for each satellte s formed by tang the receved sgnal and subtractng from t all other estmated sgnals. he frst stage of ths PIC recever structure conssts of a ban of correlators that are used to generate decson statstcs for every bt for the th satellte, Z,. hese decson statstcs then generate the estmaton of the satellte s sgnal, s (). In the next stage, as stated prevously, a new estmate for the th satellte s formed by tang the receved sgnal and substratng from t all s () such that j,, N; j. hs process may be repeated for a number of stages. Consequently, the receved sgnal at stage s for the th satellte s sgnal path s r ( s) ( s) ( r( s ( t τ ) (9) j j he decson statstc for the th navgaton data bt of satellte after s stages of nterference cancellaton s then gven as: Z ( ) τ ( s), r ( a ( t τ )cos( ωct φ ) τ () In comparson wth the SIC algorthms, the processng tme wth the PIC algorthms s greatly reduced for the large number of satelltes, but ts hardware s consderably more complcated than that of the SIC. LNA RF Front & IF-D/C Dgtzaton IF D/C ADC wth sat. wth sat. 2 wth sat. - wth sat. I&D I&D I&D I&D Regenerate sat. Regenerate sat. 2 Regenerate sat. - Stage Regenerate sat. ( ) ŝ ( ) ŝ 2 ( ) sˆ ( ) sˆ ( ) r 2 s ˆ ( ) ( ) r 2 I )( ) s ˆ ( 2 ( r ) r ( ) wth sat. wth sat. Stage 2 Fg 7. Bloc Dagram of Parallel Interference Cancellaton

7 Z. Fu and J. Wang : MAI-Mtgaton and Near-Far Resstance Archtectures 33 4 Smulatons Smulatons are made usng the MALAB Smuln pacage n the baseband, accordng to the prncple stated above. he purpose of ths smulaton s to verfy the applcablty of the algorthms and recever structures dscussed. 23 Gold codes are used, but the practcal GPS sgnal envronment and ln budget are not appled. he smulaton results on the SIC recever structure are shown n Fgs. 8, 9, and. he smulaton results on the PIC recever structure are consstent wth the results presented here. Fg Unwanted sgnal/mai sgnal level s db larger than that of SOI Fg 8. Sgnal of Interested (SOI) plus bacground nose only wthout any other MAI Fg Correlaton of SOI after usng SIC for the cancellaton of the unwanted sgnal/mai sgnal; the unwanted sgnal/mai sgnal level s db larger than that of SOI 5 Possble Enhanced Applcatons Fg.9 Unwanted sgnal/mai sgnal level s 5dB larger than that of SOI From the smulaton results shown n Fgs. 8, 9, and, t can be concluded that f the unwanted sgnal, or MAI s db larger than that of a SOI, the correlaton recever cannot properly trac the SOI PRN code. From Fg t can be seen that the SIC algorthm s applcable for MAI mtgaton or near-far resstance. Compared wth conventonal GNSS recevers, the satellte sgnal avalablty could be mproved n urban and ndoor envronments by usng the proposed GNSS recever archtectures. Furthermore, ths nd of recever archtecture could be used n pseudolte applcatons, and the ntegraton of GNSS/INS/Pseudolte, or INS/Pseudolte could mprove the system performance under a varety of poor operatonal envronments (Wang, 22). However, multpath and near-far effects reman major problems n GNSS/INS/Pseudolte, or INS/Pseudolte ntegraton for system performance mprovement. Although there have been many technques proposed to solve the near-far problem, the proposed GNSS recever archtectures are software rado archtectures whch could be a promsng approach n the future.

8 34 Journal of Global Postonng Systems 6 Concludng Remars In ths paper, the MAI-mtgaton and near-far resstant recever structures for GNSS/Galleo have been presented. he smulaton results have shown that the algorthms are applcable. Some conventonal technques as reported n Sudhr (2), for example, am to mprove the conventonal recever senstvty alone, but cannot mprove the ndoor satellte sgnal avalablty, because the conventonal structure s senstve to all the sgnals, nose, MAI and other nterference. he proposed MAI mtgaton, nearfar resstance recever structure, together wth other hgh recever senstvty technques, can mprove the ndoor and urban canyon satellte sgnal avalablty. Although the PIC recever structure s more complex than that of a conventonal recever structure, the data-floworented nature of the PIC s more sutable for mplementaton wth the DSP/FPGA, whch s a cheaper, compact, more flexble software rado approach. Further nvestgatons are requred to verfy the potental applcatons of both SIC and PIC n future GNSS recevers. Acnowledgements he authors would le to than Dr J. Cho from the School of Electrcal Engneerng and elecomm., the Unversty of New South Wales for the dscussons and hs valuable comments on ths paper. References Hen, G.W., Godet J., Issler J., Martn J.C., Lucas-Rodrguez, R. and Pratt. (2) he GALILEO Frequency Structure and Sgnal Desgn, 4 th Int. ech. Meetng of the Satellte Dvson of the U.S. Inst. of Navgaton, Salt Lae Cty, U, -4 September, aplan, D., etc. (996) Understandng GPS: Prncples and Applcatons, Artech House Publsher, Norwood, MA. Parnson, B.W., and Spler, J.J., (996) Global Postonng System: heory and applcatons, AIAA. Patel, P., and Holtman, J., (994) Analyss of a Smple Successve Interference Cancellaton Scheme n a DS/CDMA System, IEEE Journal on Selected Area n Communcatons, 2, Proas, J.G. (996 ) Dgtal Communcatons. Spler, J.J., et al., (999) Proposed New Cvl GPS Sgnals at MHz, 2 th Int. ech. Meetng of the Satellte Dvson of the U.S. Inst. of Navgaton, Nashvlle, ennessee, 4-7 September, Sudhr, N.S., et al., (2) Recever Senstvty Analyss and Result, 4 th Int. ech. Meetng of the Satellte Dvson of the U.S. Inst. of Navgaton, Salt Lae Cty, U, -4 September, Van Nee, R.D.J., (995) Multpath and Mult-ransmtter Interference n Spread Spectrum Communcaton and Navgaton Systems, Dssertaton, Delft Unversty Press. Verdu, S., (986) Mnmum probablty of error for asynchronous Gaussan Multple Access Channels, IEEE trans. Inform. heory, I-32 (): Wang, J., (22) Pseudolte Applcatons n Postonng and Navgaton: Progress and Problems, Journal of Global Postonng Systems, ():