Interoperability Test Analysis between EGNOS and MSAS SBAS Systems

Interoperablty Test Analyss between EGNOS and MSAS SBAS Systems Jorge Neto, Joaquín Cosmen, Ignaco García, GMV, S.A. Javer Ventura-Traveset, Isabel Neto, European Space Agency (ESA) Kazuak Hoshnoo, ENRI Insttute BIOGRAPHY Jorge Neto and Joaquín Cosmen receved ther MS from the Escuela Técnca Superor de Ingeneros Aeronáutcos (Unversdad Poltéctnca de Madrd, Span). Ignaco García holds MS n mathematcs from the Unversdad Complutense de Madrd, Span. They joned GMV and currently work n the GNSS Unt. In partcular, Mr. Cosmen s the manager of ths unt. GMV has a vast experence n the satellte navgaton feld beng a member of the European ndustral consortum created for EGNOS (European Geostatonary Navgaton Overlay Servce). Javer Ventura-Traveset holds a MS n Telecom. Engneerng from the Polytechnc Unv. of Catalona (Barcelona, Span, 1988); a M.S.E n Sgnal Processng by Prnceton Unversty (Prnceton, NJ) n 1992; and a PhD n Electrcal Engneerng by the Polytechnc of Turn (Italy) n 1996. Snce March 1989, he s workng at the European Space Agency (ESA) nvolved n moble, fx, earth observaton and satellte navgaton programs; he s currently Prncpal System Engneer of the EGNOS Project. Dr. Ventura-Traveset holds 4 patents and has coauthored over 80 techncal papers. He s a member of the IEEE and the Insttute of Navgaton. Isabel Neto has graduated n Electrotechncal Engneerng n Insttuto Superor Técnco (Unversdade Técnca de Lsboa, Portugal). She has joned ESA n June 1998, where she has worked wthn the EGNOS Project Team as a system Engneer. She s presently workng on the 3rd generaton moble system S-UMTS. Kazuak Hoshnoo s the Chef of the Onboard Apparatus Secton, Satellte Navgaton Dvson, Electronc Navgaton Research Insttute (ENRI), Mnstry of Transport of Japan. He has been engaged n study on the development of GNSS Integrty Channel Network and augmentaton of GNSS by a geostatonary satellte from 1992. He receved a BE n electroncs from Okayama Unversty n 1972. He joned the ENRI on Aprl n 1972 and was engaged n the research and development of moble satellte communcaton and postonng system by 1992. ABSTRACT Europe, US/Canada and Japan are currently developng ther own regonal Satellte Based Augmentaton Systems (SBAS). Although all SBAS are regonal systems, t s recognsed the necessty to establsh adequate ways for co-operaton and co-ordnaton among the dfferent SBAS provders, n order to provde SBAS nteroperablty, and, n turn, produce a more effectve mplementaton and a part of a seamless world-wde navgaton system. Although nteroperablty mples a large varety of complex ssues (such as certfcaton, standards, safety, operatons, ), n ths paper we dscuss only archtectural and techncal nteroperablty optons. In partcular, an assessment of dfferent nteroperablty optons between EGNOS and MSAS SBAS systems wll be presented. Ths assessment s based on the use of both EGNOS and MSAS respectve test beds. SBAS provders guarantee only adequate servce provson n ther nomnal servce volumes. In spte of ths, SBAS broadcast sgnals wll be avalable anywhere n ther respectve GEO footprnts. In the case of EGNOS and MSAS the GEO vsblty areas s extended to the whole Asan contnent. In that ntermedate regon (nteroperablty area), sgnals for both GEO (.e. from both SBAS) are avalable but none of the SBAS provders consders t as part of ts nomnal servce area. It means that the level of servce avalable n ths regon s not defned. Takng nto account that these ntermedate regons are not covered by any other SBAS, the possblty of provdng a mnmum servce level n the ntermedate regon by means of SBAS nteroperablty s a man concern of the nteroperablty analyses. For ths purpose, several scenaros can be conceved. These are dscussed and analysed through ths paper. The choce of one or another scenaro may mply several mportant consequences for the desgn of the dfferent SBAS, as well as standardsaton actvtes, e.g. user s recever algorthms. In ths context, an nteroperablty test actvty has been defned between EGNOS and MSAS servce provders. In partcular, real data has been smultaneously collected from the reference statons of the

EGNOS System Test Bed (ESTB) and the Japanese ENRI GNSS Test System. Usng ths real data a post-processng analyss (matched to all conceved scenaros) has been performed usng the EGNOS Early test System (ETS) platform whch ncludes prototypes of the EGNOS Central Processng Faclty (CPF) algorthms and EGNOS recever. In parallel, these scenaros have been evaluated consderng smulated data. The man objectves of these tests are to propose recommendatons for the desgn of SBAS systems and the evaluaton of performances n the nteroperablty area. Although the analyss here are orented to the EGNOS- MSAS nteroperablty case, t may be antcpated that most of the conclusons are vald for any other nteroperablty case, such as EGNOS-WAAS or MSAS- WAAS. As a contnuaton of ths analyss, a smlar test bed actvty EGNOS-WAAS s currently beng performed. INTRODUCTION There are three dfferent Satellte Based Augmentaton Systems (SBAS) to GPS (and GLONASS) currently under development: 1. In Europe, the European trpartte Group (ETG, composed of the European Unon, the European Space Agency and Eurocontrol) s n the process of developng the European Geostatonary Navgaton Overlay Servce (EGNOS). EGNOS wll cover the European Cvl Avaton Conference (ECAC) regon; 2. In the US, the Federal Avaton Admnstraton (FAA) leads the development of the Wde Area Augmentaton System (WAAS), coverng essentally contnental US (CONUS) area and Canada (Canadan WAAS CWAAS); 3. In Japan, the Japanese Cvl Avaton Bureau s mplementng Japanese MTSAT Satellte Based Augmentaton System (MSAS), whch shall cover the Flght Instrumental Regon (FIR) assocated to Japan. 90 60 30 0-30 -60 US WAAS Canadan WAAS EGNOS AOR-W AOR-E IOR MTSAT-1 POR MSAS -90-180 -120-60 0 60 120 180 Fg. 1. EGNOS, WAAS, Canadan WAAS and MSAS. Fgure 1 llustrates schematcally the world servce areas planned to be covered by these SBAS. Although all SBAS are regonal systems, t s commonly recognsed the need to establsh adequate co-operaton/co-ordnaton among SBAS provders so that ther mplementaton becomes more effectve and part of a seamless world-wde navgaton system. SBAS co-operaton s currently coordnated through the so called Interoperablty Workng Groups (IWG). Although nteroperablty mples a large varety of complex ssues (such as certfcaton, standards, safety, operatons, ), EGNOS, WAAS, CWAAS and MSAS SBAS provders have agreed on the followng lst of objectves concernng techncal nteroperablty and cooperaton among SBAS [2] and [8]: Objectve 1: Valdate SBAS performance and SIS (Sgnal n Space) consstency; Objectve 2: Defne/assess the servce level avalable n ntermedate regons between SBAS; Objectve 3: Improve ndvdual system performance though SBAS data nterchange; Objectve 4: Improve SBAS predcton capablty though SBAS data nterchange; Objectve 5: Identfy possble future mprovements. In ths paper, we wll dscuss the nteroperablty objectve 2. Some related concevable techncal scenaros are presented, together wth ther mplcatons on the SBAS and users. The prelmnary assessment of these scenaros for the EGNOS-MSAS nteroperablty case s shown. OBJECTIVE 2: DEFINE/ASSESS LEVEL OF SERVICE IN INTERMEDIATE REGIONS Although SBAS provders guarantee only adequate servce provson n ther nomnal servce volumes, SBAS broadcast sgnals wll be avalable anywhere n ther respectve GEO footprnts. In the case of EGNOS, for nstance, the EGNOS message wll be broadcast through Inmarsat AOR-E and Inmarsat IOR, whose footprnts cover together half of the globe. Ths fact, together wth the fact that EGNOS/MSAS/WAAS ntermedate regons are not covered by any other SBAS system, orgnates the debate about the possblty of provdng a mnmum servce level n the ntermedate regon by means of SBAS nteroperablty. Several scenaros may be conceved. They are dscussed hereafter where we wll talk about nteroperablty between SBAS-A and SBAS-B, and we wll consder that the mnmum desrable servce level s Non Precson Approach (NPA). A major ssue for all the nvestgated scenaros s how to guarantee the servce ntegrty out of the nomnal servce volume. An analyss of the potental concepts to cope wth t and ther mplcatons n the SBAS systems are presented after the scenaros. Scenaro 2.1: SBAS-A provdes ntegrty for the vsble GEO satelltes of SBAS-B In ths scenaro, SBAS-A provdes n the broadcast sgnal

ntegrty (and correctons) for SBAS-B GEO satelltes whch are vsble to the SBAS-A montorng network. Ths ncreases the number of montored satelltes n the ntermedate regon, whch, n turn, may ncrease the NPA servce avalablty. Consderng today s EGNOS baselne desgn, the system s dmensoned to consder the montorng of up to 8 GEO, ncludng non-egnos GEO. Thus, we may consder that ths nteroperablty scenaro s feasble f current EGNOS statons deployment s enough to montor that non- EGNOS GEO. Scenaro 2.2: Arborne recever has access to all montored satelltes from SBAS-A and SBAS-B Ths scenaro assumes that the ntegrty nformaton on the GPS satelltes generated by SBAS-A and SBAS-B may smultaneously be accessed by the avoncs at the ntermedate regon. In order to determne the navgaton soluton n ths case, the recever may use smultaneously GPS satelltes that are montored by SBAS-A and GPS satelltes that are montored by SBAS-B. Certanly, then, the number of GPS satelltes that are globally montored may be large enough so that wth no extra reference statons, NPA servce level s avalable n the ntermedate regons. The mplementaton of ths scenaro s as follows: SBAS- A and SBAS-B provde ther own broadcast navgaton messages, whch can be used by the navgaton recever smultaneously. Correctons and ntegrty nformaton from the navgaton messages of two dfferent SBAS can safely be used smultaneously n the determnaton of the user poston (see fgure 2). staton sends to SBAS-A master staton correctons and ntegrty nformaton on GPS satelltes whch are not vsble to SBAS-A. SBAS-A then consders ths nformaton n the generaton of ts navgaton sgnal (addng the ntegrty nformaton on those non-vsble satelltes). Although n terms of fnal user performance the two scenaros above (2.2 and 2.3) could be equvalent, the mplcatons of those are very dfferent. Indeed, scenaro 2.2 mples a rather lmted (f any) nterface between the two SBAS, whle t mples the need to defne a mnmum avoncs requrement where ths smultaneous use of SBAS sgnals s contemplated (today RTCA MOPS [3] does not consder clearly ths case). Scenaro 2.3, nstead, could eventually be completely transparent to the user recever (whch s actually gettng the sgnal from only one SBAS) but would mply a very complex nterface between SBAS wth many safety mplcatons. Based on these consderatons, scenaro 2.2 s a pror more attractve. Scenaro 2.4: Installng own montorng statons by each SBAS provder and provdng dual servce In ths case, SBAS-A and SBAS-B systems mplement some addtonal reference statons (n adequate stes) n such a way that both SBAS provde servce n the ntermedate SBAS regon ndependently. The nteroperablty, n ths case, may consst only n the provson of servce redundancy, allowng the user to jump to the alternate SBAS sgnal n case of contnuty problems wth the current SBAS sgnal n use. SBAS nteroperablty n ths case could mprove both contnuty and avalablty of servce n the ntermedate regons. In fact, ths scenaro can be consdered as the expanson of the servce provded by each SBAS to the nteroperablty regon. Extendng the ntegrty data outsde SBAS nomnal servce volumes (UDRE out of zone degradaton) Fg. 2. Scenaro 2.2. Scenaro 2.3: Arborne recever has access to all montored satelltes from SBAS-A and SBAS-B through a sngle SIS The concept behnd ths scenaro s smlar to the prevous one, but the mplementaton s completely dfferent: SBAS master statons do provde to each other the relevant nformaton, whch s ntroduced n each SBAS message ndependently. For nstance SBAS-B master Any of the scenaros lnked to Objectve 2 assumes that the ntegrty nformaton provded by the SBAS s avalable n a larger area than the nomnal servce volume defnton. In the extreme, we may assume that the ntegrty nformaton should be vald anywhere n the GEO footprnt assocated to a gven SBAS. Assumng agan that Objectve 2 nteroperablty s lmted to the provson of En-route to NPA servce, the ssue s then lnked to: 1) the valdty of the UDRE bounds (valdty of satelltes correctons ntegrty) n that extended area and more mportantly 2) the valdty of the Horzontal Protecton Level (HPL) (valdty of the user navgaton ntegrty). Related to that, several optons may be consdered: Opton 1: UDRE s computed consderng not only the target servce area but also the whole GEO footprnt or the nteroperablty area,.e. SBAS requrements are modfed so that UDRE bounds are vald everywhere wthn

footprnts. The obvous mplcaton of that approach s that the UDRE values wll be ncreased, whch, n turn, could affect the avalablty n the nomnal servce volume of Precson Approach. Opton 2: Formally demonstrate that keepng UDRE values (as determned for the nomnal servce volumes) the worst-case error n the GEO footprnts s always small relatve to the allowances ncluded n the HPL equaton assocated to NPA (these allowances are essentally to account for possble onospherc errors). The analyss of ths scenaro requres the understandng and assessment of possble pathologcal cases,.e., those n whch based on geometry vsblty consderatons, a satellte error n a partcular dmenson occurs n such a way that t can not be observed by the montorng network but stll affect the user. Opton 3: Apply a degradaton factor for En-route to NPA to account for the possble UDRE degradaton. Ideally, ths degradaton factor shall be appled outsde the SBAS nomnal servce volume (not to mpact PA) and should have no avalablty mpact. Ths factor could be appled by the recever wthout the need to be transmtted n the SBAS message. Alternatvely, the factor could be ncluded n the SBAS SIS (e.g. an adaptaton of the former MOPS message 27 has currently been proposed for that). It s worth mentonng that the ssue of ntegrty broadcast outsde the servce volumes s not only an ssue related to ths nteroperablty scenaro but a more global one. Indeed, even f SBAS do lmt ther commtted servce provson to ther nomnal servce volumes, the broadcast sgnals wll be avalable anywhere n the footprnts of the SBAS GEO satelltes. Thus, unless provsons are standardsed to nform the recevers whether they are or not n the servce volume of a partcular SBAS (and so whether they are or not allowed to use the SBAS broadcast sgnal), nothng prevents a recever to access and process these data, wth all the assocated ntegrty related problems. ASSESSMENT OF THE INTEROPERABILITY SCENARIOS ASSOCIATED TO THE OBJECTIVE 2 In order to assess the nteroperablty scenaros, they have been mplemented n the EGNOS Early Test System (ETS) platform [1]. The ETS s a functonal end-to-end EGNOS prototype manly addressed to nvestgate the EGNOS performance at user level n terms of accuracy, ntegrty and avalablty. It mplements major EGNOS functons, payng specal attenton to the dfferent algorthms that wll be mp lemented n the Central Processng Faclty (CPF) of EGNOS. Ths faclty can be easly adapted to analyse the performance assocated to any other SBAS. In partcular, the nteroperablty case consdered s the one between EGNOS [6] and MSAS [7] systems. Two dfferent analyses have been carred out, one based on real data and the other based on smulated data. Real data analyss SBAS ground segment data has been provded by EGNOS and MSAS test beds: EGNOS System Test Bed ESTB [4] and ENRI GNSS Test System respectvely. Data analysed corresponds to the day 15 th of February 1999. Ths data has been processed n the ETS faclty, n order to generate the emulated SIS for EGNOS and MSAS. Fgure 3 shows the statons consdered n the analyss: fve of them correspond to EGNOS and the other fve to MSAS. For the user segment, fourteen IGS statons located n the nteroperablty area have been consdered (fgure 4). Fg. 3. EGNOS and MSAS ground segment for real data. Fg. 4. User ground segment (IGS statons) for real data. Smulated data analyss The second assessment of the nteroperablty scenaros s performed wth smulated data. EGNOS and MSAS ground segments (fgures 5 and 6) and user segment have been realstcally smulated. Smulaton tool consdered s the EGNOS End-to-End Smulator, EETES [5]. Fg. 5. EGNOS ground segment for smulated data.

Fg. 8. Maxmsaton of the mnmum number of satelltes: SBAS-A fve, SBAS-B fve. Fg. 6. MSAS ground segment for smulated data. IMPLEMENTATION ASPECTS Each nteroperablty scenaro allows dfferent techncal solutons for the system mplementaton. The descrpton of those dfferent optons s presented for each scenaro. Due to the lmtatons of the ETS, Scenaro 2.1 has not been mplemented. Scenaro 2.2 In ths scenaro, users access smultaneously to both SBAS SIS, and combne ther nformaton n a sngle navgaton soluton. Regardng ths scenaro, two techncal ssues need to be consoldated: 1. The crtera for the selecton of correctons for those satelltes whch are smultaneously montored by both SBAS. 2. The effect of dfferent SBAS reference tme for satellte clock correctons, whch may degrade user s performance. For the frst ssue (SBAS selecton for satelltes doubly montored), three dfferent approaches have been consdered (fgures 7 and 8 llustrate the two frst approaches): 1. Maxmse the maxmum number of satelltes montored by the same SBAS. 2. Maxmse the mnmum number of satelltes montored by the same SBAS. 3. Select the SBAS provdng the mnmum UDRE. For the second ssue (tme offset between SBAS), two approaches have been analysed: 1. Tme offset s not estmated. In ths case, standard navgaton algorthm (four unknowns: user s locaton and recever clock bas) s consdered. 2. Tme offset s estmated. In consequence, navgaton algorthm s modfed n order to estmate fve unknowns: user s locaton, recever clock bas w.r.t. SBAS-A and the tme offset between SBAS-A and SBAS-B. In the second approach, GPS measurement equaton presented n the Appendx E of MOPS [3] s modfed as follows: G = y = G x + ε cos Az cos El snaz [ cos El snel 1 ζ ] where x s a fve dmensonal poston vector (north, east, up, clock w.r.t. SBAS-A and clock of SBAS-B w.r.t. SBAS-A) and ζ 0 = 1 Scenaro 2.3 f f SBAS - A correcton s are appled SBAS - Bcorrecton s are appled In ths scenaro, users access smultaneously only to one SBAS SIS, but ths SIS ncludes the nformaton from both SBAS. It means that the nformaton generated by SBAS-B s used by SBAS-A to complete ts SIS wth those satelltes not montored by A but montored by B. As t happens n the scenaro 2.2, t s requred to decde f the tme offset between SBAS s gong to be estmated or not. For ths scenaro, the followng approaches have been analysed: Fg. 7. Maxmsaton of the maxmum number of satelltes: SBAS-A seven, SBAS-B three. 1. Tme offset s not estmated. Ths case s smlar (from performance pont of vew) to the one analysed n scenaro 2.2. 2. The ground segment estmates tme offset. Once ths offset has been estmated, the broadcast clock nformaton s corrected n the ground segment to elmnate ths term. In consequence, from user s pont

of vew, all the satelltes are smlar and standard navgaton algorthm s consdered. In the second approach, the ground segment estmates the tme offset (B A-B ) as follows: B p ( ba, bb ) = A B = 1, where b A and b B are the satellte clock correctons for SBAS-A and B and =1... p are those satelltes that are smultaneously montored by both SBAS. Scenaro 2.4 In ths scenaro, both SBAS mplement addtonal statons n order to provde ndependently the requred navgaton servce n the nteroperablty area. In consequence, users have access to two navgaton solutons, one per SBAS. Regardng ths scenaro, two ssues need to be consoldated: 1. The locaton of the addtonal montorng statons. They can be co-located wth statons belongng to the other SBAS (n order to reduce cost as e.g. common securty, mantenance, surveyng, etc.), or not (n order to optmse performance through an adequate deployment). 2. The management of the two possble solutons, e.g. the crtera for swtchng from one soluton to the other or the use of one soluton to montor the other. The tests have been performed consderng four addtonal statons per SBAS. In real data analyss, only co-located case has been analysed. In the smulated data analyss, three dfferent cases have been consdered: four co-located statons, three co-located plus a new one and two colocated plus two new ones. For the management of the two possble solutons, t s proposed to compute both smultaneously and then: If only one soluton s avalable (HPL < HAL), ths soluton s selected. If both solutons are avalable, two approaches are consdered: 1. Select the same soluton than n the prevous epoch. 2. Cross-check both solutons,.e. check the coherence between the estmated user s locatons and the assocated protecton levels. Ths technque could mprove ntegrty but decreasng avalablty. RESULTS The assessment of the nteroperablty scenaros has been based on the executon of a set of tests n the ETS p platform, consderng both real and smulated data. For each test, the followng user s performances have been evaluated: Horzontal accuracy: 95 th percentle of the horzontal postonng error dstrbuton. Avalablty: relatve frequency of the number of cases where the NPA navgaton servce was avalable (HPL < HAL=556m). Integrty: relatve frequency of the number of cases where the NPA navgaton servce was declared avalable and postonng errors were below ther correspondng protecton levels (horzontal error < HPL). As t has been commented before, NPA phase of flght has been consdered as the reference for all the tests. Man results and conclusons derved from the tests are presented hereafter. Reference scenaro ( do nothng ) In order to compare results, a reference scenaro has been proposed. It assumes that each SBAS (EGNOS and MSAS) s provdng the nomnal navgaton servce n ther respectve servce areas. There s not any specal provson regardng those users located n the nteroperablty area. In spte of ths, users located outsde these servce areas are able to use EGNOS or MSAS nformaton. In ths case, the degradaton of user performance for those users located outsde the servce areas (.e. n the nteroperablty areas) can be observed. Ths degradaton ncreases when the users are located far from the respectve servce areas. The degradaton of performance s clearly assocated to the reducton of the number of montored satelltes for these users. Fgures 9 and 10 show the horzontal accuracy assocated to the smulated data tests. Table 1 presents the average of the performance fgures obtaned for all the users. In spte of these averaged values can not be consdered as the actual system performances, they can be useful for comparson purposes. Fg. 9. Horzontal accuracy (meters) assocated to MSAS SIS, smulated data.

Fgure 11 shows the horzontal accuracy assocated to the smulated data tests. Table 2 presents the mean values assocated to the real (wth a lmted number of statons) and smulated data users. Fg. 10. Horzontal accuracy (meters) assocated to EGNOS SIS, smulated data. SIS Horzontal accuracy (m) Avalablty MSAS 87.9 66.0 EGNOS 139.9 51.0 MSAS 55.4 69.0 EGNOS 15.3 93.4 Table 1. Mean user s performance for real (shadowed) and smulated data. Scenaro 2.2 Scenaro Horzontal accuracy (m) Avalablty Integrty 2.2 8.6 94.9 100.0 2.2 9.2 99.9 100.0 Table 2. Mean user s performance for real (shadowed) and smulated data. Scenaro 2.3 After the evaluaton of the approaches consdered for ths scenaro, t s recommended to estmate the tme offset between SBAS-A and SBAS-B. Ths approach reduces the postonng errors. Addtonally, t has the advantage of not reducng the number of redundant measurements at user level, as users do not requre to estmate that offset. Fgure 12 shows the horzontal accuracy assocated to the smulated data tests. Table 3 presents the mean values assocated to the real (wth a lmted number of statons) and smulated data users. After the evaluaton of the dfferent approaches consdered for ths scenaro, the followng recommendatons are proposed: It s recommended to estmate the tme offset between SBAS-A and SBAS-B. Otherwse, postonng errors ncrease up to unacceptable levels. As far as each SBAS tme s wthn 50 nanoseconds (15 meters) of GPS tme (Appendx A of MOPS [3]), ths result could have been expected. The combnaton of satelltes montored by dfferent SBAS degrades performances: t s requred to estmate an addtonal parameter, reducng the number of redundant measurements. It has mplcatons on avalablty and accuracy performances. Under these condtons, t s recommended to maxmse the maxmum number of satelltes montored by the same SBAS. Ths approach reduces the number of cases where both SBAS are used smultaneously. Fg. 12. Horzontal accuracy (meters) n scenaro 2.3. Scenaro Horzontal accuracy (m) Avalablty Integrty 2.3 9.8 99.2 99.8 2.3 8.3 100.0 100.0 Table 3. Mean user s performance for real (shadowed) and smulated data. Scenaro 2.4 After the evaluaton of the dfferent approaches consdered for ths scenaro, the followng recommendatons are proposed: Fg. 11. Horzontal accuracy (meters) n scenaro 2.2. The locaton of the addtonal statons seems to have a small nfluence n the performances. Each SBAS could take the advantage of the exstng statons of the other SBAS n order to share costs. No specal mprovements are shown by the crosscheck technque. The selecton of one of the avalable solutons s a smple and effectve selecton method.

Fgures 13 and 14 show the horzontal accuracy assocated to the smulated data tests, for EGNOS and MSAS SIS. Table 4 presents the mean values assocated to the real (wth a lmted number of statons) and smulated data users. Fgure 15 presents the horzontal protecton levels (HPL) versus horzontal postonng errors for real data when the users are allowed to select, at each epoch, one of the avalable solutons (EGNOS or MSAS). In ths case, avalablty performance s 99.2 % and horzontal accuracy s 6.9 meters (100 % and 8.2 meters respectvely for smulated data). UDRE out of zone degradaton Fg. 13. Horzontal accuracy (meters) n scenaro 2.4 for EGNOS (four addtonal statons). The results prevously presented have been performed consderng the opton 3 (see related secton above) of UDRE out of zone degradaton: the extenson of the UDRE for out of zone users s performed at user level consderng a constant but SBAS dependant factor. Addtonal tests have been performed consderng the other two optons: opton 1, computaton of UDRE to bound errors n both, servce and nteroperablty areas; opton 2, no expanson,.e. usng everywhere the UDRE values computed for the servce area. Whle optons 1 and 3 should provde the requred ntegrty performance (consderng approprate measurements have been taken to guarantee these performances), tests for opton 2 wll be useful to evaluate f that ntegrty performance s acheved, and only n that case measured avalablty performance wll be meanngful. Fg. 14. Horzontal accuracy (meters) n scenaro 2.4 for MSAS (four addtonal statons). SIS Horzontal accuracy (m) Avalablty Integrty MSAS 7.5 97.8 100.0 EGNOS 7.0 98.7 100.0 MSAS 8.8 99.7 100.0 EGNOS 8.2 100.0 100.0 Table 4. Mean user s performance for real (shadowed) and smulated data. Table 5 presents the user s performances assocated to the real and smulated users for the dfferent UDRE out of zone degradaton optons. Opton 2 has only been analysed wth real data. Scenaro consdered s the 2.4, selectng one of the avalable soluton at each epoch. UDRE opton Horzontal accuracy (m) Avalablty Integrty 1 7.0 99.4 100.0 2 6.9 99.4 100.0 3 6.9 99.2 100.0 1 8.2 100.0 100.0 3 8.2 100.0 100.0 Table 5. Mean user s performance for real (shadowed) and smulated data. Takng nto account the results of these tests, the followng ponts can be concluded: Fg. 15. HPL versus horzontal postonng errors. For NPA phase of flght, UDRE represents a small contrbuton n the protecton level computaton. It s deduced from the fact that avalablty performance are slghtly nfluenced by the dfferent approaches consdered. It means that the major part of the error budget n the protecton levels s absorbed by the other terms, especally onosphere. Therefore, t does not seem to be crtcal to refne UDRE values to obtan a very adjusted orbtal and clock error budget. The same effect can be observed from ntegrty pont of vew. For accuracy, effect s mnmum, as the

unque effect s shown through the weghtng factors n the weghted least squares algorthm. The computaton of UDRE to bound errors n both, servce and nteroperablty areas (opton 1) could be crtcal for PA users, whch are located n the orgnal servce area of each SBAS. In ths case, the margns are qute reduced and the nfluence of UDRE term n the protecton levels s mportant. Therefore, ths opton s n prncple rejected n order to avod any degradaton n the performance of the nomnal servce volume. Opton 2 s not recommended as advantages n avalablty are mnor but effect n ntegrty could be crtcal n the case of extreme condtons (e.g. worst case onosphere), even though, ntegrty s guaranteed for the analysed scenaros. In consequence, our recommendaton s to consder a degradaton factor at user level (opton 3): t can be used to guarantee ntegrty under extreme condtons, wthout affectng sgnfcantly avalablty under normal condtons. Ths degradaton factor could be ether fxed and mplemented n the user (as n our tests) or broadcast by the SBAS ground segments (e.g. by allowng a redefnton of MOPS message 27). It s worthwhle to menton here that the expanson of the servce to the nteroperablty area mples another modfcaton n the SBAS desgn, the UDRE computaton functon. The pont here s that UDRE (consdered as the bound of the orbt and clock error for the worst user n the servce area) has no sense when the satellte s not vsble n the servce area. It could happen (e.g. n the scenaro 2.4) that a satellte can be successfully montored by the SBAS but t s only vsble n the nteroperablty area. Our proposal for ths ssue has been to estmate the UDRE for ths type of satelltes as the bound of the orbt and clock error for the worst user n the nteroperablty area. CONCLUSIONS In ths paper, objectve 2 of SBAS nteroperablty,.e. to defne/assess the servce level avalable n the ntermedate regons between SBAS, has been analysed n some detal. In partcular, three related nteroperablty scenaros have been compared consderng real and smulated data for the case of EGNOS and MSAS nteroperablty. Man conclusons are outlned n the next paragraphs: 1. Takng nto account the lmtatons of the real and smulated data analysed, numercal values should not be consdered as absolute but only relatves ones (as comparson among dfferent scenaros). It s mportant to observe that the results for both types of analyses (real and smulated data) are coherent: the relatve postons of each scenaro wth respect to others s the same and, therefore, both type of analyses support the conclusons; 2. As a general concluson for all the scenaros, nteroperablty may requre modfcatons on current SBAS desgns. These modfcatons come from the UDRE out of zone degradaton. There s a need of common defnton and standardsaton of UDRE out of zone wth the objectve of provdng ntegrty even out of the nomnal SBAS servce areas. A redefnton of MOPS message 27 s currently under dscusson for ths purpose, and t s subject to analyss on a dedcated EGNOS-WAAS nteroperablty actvty. 3. Accordng to our results, the recommended nteroperablty scenaro s the 2.4 one,.e. each SBAS mplements some addtonal reference statons n such a way that both SBAS provde navgaton servce n the ntermedate regon ndependently from each other. Each SBAS s responsble for provdng the requred navgaton servce. Addtonal (to the one above stated) mplcatons on SBAS and user desgn are more quanttatve than qualtatve: Addtonal statons are needed. They could be colocated wth statons belongng the other SBAS or a combnaton of co-located exstng statons and ndependent new statons. Addtonal SBAS computaton power,.e. CPU, s requred. For EGNOS ths s not a crtcal problem because ts Central Processng Faclty (CPF) s dmensoned up to 60 RIMS (when current baselne ncludes 39). Addtonal nformaton shall be ncluded n the broadcast SIS. It does not seem to be crtcal, as the addtonal messages to broadcast are the long term correctons assocated to the new satelltes montored. The typcal refresh rate of ths type of message s 2 mnutes. Addtonal user computatons power could be requred, n order to execute the navgaton algorthm twce per second (one for each SBAS). As t can be observed, these modfcatons do not mply any techncal nnovaton on SBAS or user desgn. Another major advantage comes from the certfcaton pont of vew as each SBAS s fully responsble for the provson of a servce n the nteroperablty area totally ndependent from the other SBAS. 4. Results assocated to scenaro 2.2 are slghtly worse than those of scenaro 2.3 and 2.4. The effect of the tme offset between SBAS (and the subsequent loss of a measurement for ts estmaton) has crtcal consequences on the performances. It s mportant to hghlght that the real data tests have been performed wth a lmted number of statons. Wth a more realstc stuaton n the smulated data tests (referred to the number and geometrcal dstrbuton of RIMS), avalablty results are mproved but stll below 100%. Ths s especally true n the proxmty of MSAS servce area, due to the low number of statons n MSAS ground segment (as compared to EGNOS

ground segment). From a conceptual pont of vew, ths scenaro could have mportant mplcatons on user desgn, as t s the modfcaton of standard navgaton algorthm to nclude the estmaton of the SBAS tme offset. One possble soluton to the SBAS tme offset drawback s to nclude an addtonal requrement to each SBAS to reach a synchronsaton closer to GPS tme. In ths case, users would not need to estmate the offset between SBAS, and the accuracy degradaton would be mnmsed, takng nto account that the SBAS tme offset s below a certan lmt 5. Scenaro 2.3 provdes better performances than scenaro 2.2, as the users do not lose one measurement for offset estmaton: avalablty and accuracy ncrease. In any case, ths scenaro 2.3 mples some mportant modfcatons on the SBAS desgn, as t s the case of the requred lnk between SBAS. Each SBAS shall be able to send to the other orbt and clock correctons besdes UDRE values for all the montored satelltes at each second. Addtonally, latency n the delvery of ths nformaton should be mnmsed. From a practcal pont of vew, ths s consdered as a very dffcult to mplement soluton. A possble alternatve for ths scenaro could be based on the followng approach: each SBAS receves the nformaton from the other through ts SIS, estmates the tme offset and broadcasts exclusvely ths tme offset n, say, MOPS message type 12. 6. Numercal performances obtaned are qute promsng and t s antcpated that t s possble to provde an NPA servce level n the nteroperablty area. Despte the lmtatons n the number of statons consdered, avalablty fgures for NPA n the EGNOS-MSAS ntermedate regon are of the order of 99.9% (whch, n turn, result n avalablty fgures of the order of 99.999% when RAIM s also consdered as a backup) for scenaros 2.2, 2.3 and 2.4. Scenaro 2.4 provdes a homogeneous NPA performance dstrbuton for all the users. Whle numercal avalablty values have been analysed from a relatve pont of vew, t s truth that accuracy can be checked from an absolute pont of vew. Global accuracy values (horzontal around 7 meters for real data tests and 8 meters for smulated data tests, 95%) can be consdered as excellent (takng nto the characterstcs of the data analysed). It s mportant to hghlght that results could mprove f, for nstance, GEO rangng were ncluded. ACKNOWLEDGEMENTS The work descrbed n ths paper was done under ESA contract. The authors want to acknowledge the support provded by ESA and ENRI staff for the tmely provson of data and the frutful techncal dscussons related to the work here descrbed. REFERENCES 1. Assessment of EGNOS System and Performance: Early Test System, J. Neto, M.A. Molna, M.M. Romay, J. Cosmen, M.L. de Mateo, R. Román, L. Andrada; Proceedngs of ION GPS 97. 2. A Techncal Revew of SBAS Interoperablty Issues from the EGNOS Perspectve ; J. Ventura-Traveset, C.F. Garrga, I. Neto, J.M. Peplu, E. Sales, X. Derambure; Proceedngs of GNSS 98. 3. Mnmum Operatonal Performance Standards (MOPS) for Global Postonng System /Wde Area Augmentaton System Arborne Equpment, RTCA/DO-229, Change 1 and RTCA/DO-229A. 4. EGNOS System Test Bed Archtecture ; P. Razonvlle, R. Hansen, P. Goun, J.M. Gaubert, N. Zarraoa; Proceedngs of GNSS 98. 5. Analyss of EGNOS Crtcal Desgn Issues Through Prototypng: UDRE Computaton and GEO Orbt Determnaton ; H. Boomkamp, J. Cosmen, J.A. de la Fuente, J. Neto, G. Salgado, E. Sardón, M. Toledo, P. Van Nftrk, N. Zarraoa; Proceedngs of GNSS 98. 6. EGNOS Project Status Overvew, J.Benedcto, P.Mchel and J.Ventura-Traveset, Ar&pace Europe Journal (Elselver), No.1, Vol. 1, pp.58-64, January- February, 1999. 7. MSAS (MTSAT Satellte-based Augmentaton System) Project Status ; A. Shmamura; Proceedngs of GNSS 98. 8. STID: SBAS Techncal Interface Document (STID) for Interoperablty, J.Ventura-Traveset et al., EGNOS, WAAAS, Canadan WAAS and MSAS jontly produced document at Interoperablty Workng Group, March 22, 1999. 7. Although the analyss here are orented to the EGNOS-MSAS nteroperablty case, t may be antcpated that most of the conclusons are vald for any other nteroperablty case, such as EGNOS- WAAS or MSAS-WAAS. As a contnuaton of ths analyss, a smlar test bed actvty EGNOS-WAAS s currently beng performed.