GNSS Integrity for Railway Transportation

Transcription

1 GSS Integrty for Ralway Transportaton M. Jonáš Faculty of Electrcal Engneerng an Informatcs, Unversty of Parubce, Parubce, Czech Republc * Corresponng author: marek.jonas.stuent@upce.cz DOI: /v ABSTRACT: Ths artcle contans an analyss of GSS (Global avgaton Satellte System) ntegrty from a vewpont of ralway transportaton. The ntegrty concept of functonal EGOS (European Geostatonary avgaton Overlay Servce) system s explane n etal an also the ntegrty mechansm of the future Galleo system s brefly outlne. In orer to verfy the theoretcal conclusons, statc measurements by means of EGOS recever n a safety moe have been performe. Selecte expermental results are scusse. KEY WORDS: GSS ntegrty, ralway, EGOS, DGPS, GPS, Galleo. 1 MOTIVATIO Before a satellte navgaton system such as EGOS or Galleo can be use n ralway safetyrelate applcatons, t s necessary to perform a rsk analyss of the whole ralway safetyrelate system an specfy ts safety ntegrty an epenablty requrements. It s manatory to perform the rsk analyss accorng to the ralway safety stanars (E 5016, E 5019, etc.). For ths reason, EGOS epenablty attrbutes as qualty measures of one of subsystems must be etermne - accorng to the ralway safety concept. DESCRIPTIO OF GSS ITEGRITY AD AVAILABILITY FOR RAILWAY EVIROMET It s well known that contons for the applcaton of GSS n avaton an on ralway are very fferent. Ths s manly ue to SIS (Sgnal-In-Space) shaowng by fferent objects along the ralway lne or by a lanscape profle, an also ue to the more emanng requrements for safety an epenablty on ralway. The total ntegrty of GSS postonng can be nfluence by errors n a space segment, errors ue to SIS propagaton effects n the atmosphere, errors ue to multpath effects an fnally by errors ue to potental falures n the user recever. The error sources wth a potental mpact on SIS ntegrty an effects of ralway envronment are epcte n Fg

2 .1 SIS ntegrty Ths means the ntegrty of SIS transmtte by satelltes. Integrty s a measure of the trust whch can be place n the correctness of the nformaton supple by the system. Integrty nclues the ablty of the system to alert the user when the system shoul not be use for the ntene operaton. At ths tme, the system EGOS has been certfe for use n avonc safety crtcal applcatons snce December 010 (Safety of Lfe Servce). Fgure 1: Effects on GSS ntegrty an SIS avalablty.. SIS avalablty SIS avalablty s affecte by fferent contons along the ralway lnes. Ths part can be ve nto three basc subparts: sngle lne, ouble or multple lne, ralway staton (many rals). The contons for GSS sgnal recepton are fferent n each of these cases. The poston can be etermne only n a 1D oman at a sngle lne. Ths leas to the use of smpler algorthms for ntegrty verfcaton. At a sngle lne there are often worse contons for GSS sgnal recepton, ue to brges, nearby bulngs, trees an forests along the lne. The poston s etermne n a D oman to stngush 184

3 on whch of the two or more parallel lnes the tran s stuate n case of a ouble or multple lne. There wll be sgnfcantly better contons for GSS sgnal recepton. In the case of a ralway staton, where there are usually many parallel lnes, there are frequently the best contons for GSS sgnal recepton. On the other han, t wll be harest to ece on whch of the many parallel lnes the tran s stuate. Durng the GSS sgnal recepton on earth groun there s also sturbng by multpath..3 Poston ntegrty Poston ntegrty s a measure of the trust whch can be place n the correctness of the estmate poston. Poston ntegrty s effecte by SIS avalablty an multpath. Poston ntegrty can be mprove by ang nertal sensors (IS) such as an oometer, accelerometer, gyroscope an mcrowave Doppler speeometer. Data from IS can be fuse by a Kalman flter an projecte to the map of ralway lnes. It s thereby possble to check the poston ntegrty by means of usng maps. The mathematcal equatons whch can be use were presente n [7]. Also autonomous ntegrty montorng can be use to ncrease poston ntegrty; ths means that the system compares the estmate postonal error (represente by horzontal stanar evaton estmate by the recever) wth the current level of the horzontal protecton level (HPL) whch wll be explane further. 3 CURRET EGOS ITEGRITY COCEPT The EGOS ntegrty concept s escrbe n stanar DO-9D. The EGOS system has three satelltes whch prove nformaton about system ntegrty. The EGOS system sgnal s avalable throughout the whole of the European terrtory. The EGOS sgnal s partally avalable even n Asa an Afrca, as s epcte n fgure. The ots n the graph represent onospherc gr ponts (ponts where onospherc correctons are avalable). The ots whch are marke by re crcles represent the terrtory where sgnal s theoretcally avalable. Ths graph was generate by Pegasus software, whch s beng evelope by Eurocontrol for EGOS an Galleo valaton tests. Fgure: Terrtory where EGOS SIS s avalable. 185

4 Wthn the Safety of Lfe Servce there are two navgaton moes an ther relate maxmal angerous msse etecton falure rates n a fault-free case are as follows: 7 Precson Approach (PA): PA 110 / 150 s (planes approach) 7 on-precson Approach (PA): / hour (urng the plane flght) PA 1 In both these moes a safety-relate groun recever computes the horzontal an vertcal protecton levels (HPL, VPL) from the ata obtane n each epoch. For computng protecton levels the system uses only ata from satelltes whch are consere to be healthy (fault-free case). The current level of HPL ncates the area (a crcle aroun the current user poston) n whch the above-mentone falure rates are fulflle. From the pont of vew of ralway transportaton HPL s mportant, because we preetermne the poston n the horzontal plane [1, ]. The essental nput quanttes for HPL computaton are: geometry between GPS satelltes an the user (elevaton El an azmuth Az of the th observe satellte), user fferental range error (varance tropospherc error (varance, flt ), gr onospherc vertcal error (varance,uire ), ).,tropo ), an the error of arborne recever (varance,ar The accuracy of these parameters can be reuce by an ephemers error (the fference between the expecte an actual orbtal poston of a GPS satellte) an a satellte clock error. HPL equatons: K HPL K H, PA H, PA major major major s the sem-major axs of error ellpse an s calculate as: major east north east north E from projecton matrx S: S east E EU ET E north U T EU U U UT UT T ET T T G W G 1 186

5 where: east s east, 1 = varance of moel strbuton that overbouns the true error strbuton n the east axs north s north, 1 = varance of moel strbuton that overbouns the true error strbuton n the north axs E east, north, 1 U s u, 1 s s = covarance of moel strbuton n the east an north axs = varance of moel strbuton that overbouns the true error strbuton n the vertcal axs th an row of the geometry matrx G s efne wth elevaton El an the azmuth Az of the th observe satellte as: G cos El sn Az cos El cos Az sn El 1 Matrx W s moele uner the assumpton of uncorrelate measurements characterze by the varance for the observe satellte as follows:, flt, UIRE, ar, tropo A more etale vew on the computaton of ntegrty parameters s avalable n self stanar DO-9D [1]. The key thng s the ervaton of constant K H, PA, t was orgnally chosen to be consstent wth certan assumptons on the strbuton of poston error an on correlaton tme error. It s relate to the probablty of msse etecton ( Pm ) of msleang nformaton (MI), where MI means that horzontal poston error (HPE) s larger than HPL. Pm HPL 10 n X n PA x Where 10 s the ntegrty requrement for ths operaton (n our case falure rate PA ), an n s the number of nepenent samples per operaton. 187

6 The number of nepenent samples per tme unt n EGOS, base on onospherc correctons, 360 s was aopte as a reasonable assumpton to ensure nepenence [5]. Pm HPL PA 9 PA 510 per sample K factor scales the varance to a level compatble wth the ntegrty requrement. In the case of HPL, snce the protecton has to be b-mensonal, K s etermne from a Raylegh strbuton. Factor K s rectly calculate from the knowlege of the cumulatve strbuton functon (cf) of the relevant statstcal law: K H PA Raylegh cf Pm Raylegh cf HPL So constant K, was set as 6.18 base on the assumpton that the ecorrelaton tme H PA of EGOS errors s 360 s. However, an analyss of ths assumpton was one n [3]. It therefore seems that presently ths assumpton has not been fulflle by the EGOS system [1, 3, 4]. 4 GALILEO ITEGRITY COCEPT In the upcomng satellte navgaton system Galleo, all satelltes wll broacast ntegrty nformaton, so t wll be avalable worlwe. However the Galleo ntegrty mechansm wll be fferent from EGOS ntegrty one. Unlke the EGOS concept, where the system computes horzontal an vertcal lmts for a gven fxe ntegrty rsk, a Galleo recever wll compute ntegrty rsk for the user efnng horzontal an vertcal level (HAL horzontal alert lmt, VAL vertcal alert lmt). Thus a Galleo ntegrty rsk epens on the user specfe alarm lmt of nterest [5]. The relaton between both ntegrty concepts an possbltes of usng nformaton from both ntegrty concepts was analyze for example n [6]. 5 PRACTICAL EXPERIMETS The EGOS system s certfcate for use n avonc safety crtcal applcatons, but n ralway applcatons t s necessary to analyze the real performance of the system on the earth groun an to valate the fulfllment of EGOS parameters on earth groun, an so to verfy the theoretcal propertes of the system. Data collecton was carre out wth the groun safety-relate GSS recever PolaRx3. Data collecton was performe by means of the current avalable EGOS system over three ays, from 10 to 13 May 011 n Parubce n Czech Republc. Fg. 3 splays the poston error (HPE) an HPL. HPL s compute by the recever accorng to the above-mentone equatons. In the graph t seems 188

7 that there are some relatvely bg jumps n the tme behavor of HPL, whle the level of HPE s approxmately constant. Fgure3: Horzontal poston error an horzontal protecton level. Fg.4 shows the measure poston ponts (green ponts) an the true poston of the antenna, whch s marke by the re trangle. From the graph t seems that the varance n horzontal an vertcal recton s approxmately 3 meters. The mean value of measure ponts s marke by the re star. The re crcle wth the center at the antenna poston means crcular error probablty (CEP). CEP means the raus of a crcle whch contans 50% of poston ponts. It s compute accorng to the equaton: CEP( 50) 0,588( RMS E RMS ) where: RMS E 1 Delta E, RMS 1 Delta, Delta E, an Delta are evatons between the true an measure poston n east an north rectons. RMS E, an RMS are corresponng mean square errors n east an north rectons. 189

8 The raus of the smaller blue crcle s the stance root mean square (RMS), an the larger blue crcle ha a raus RMS. It s compute per equatons: RMS RMS RMS RMS RMS E Parameters CEP, RMS, an RMS express D accuracy of GSS recever. Fgure4: Measure poston ponts. The number of avalable GPS satelltes urng ata collecton was between 5 an 1. The curve of numbers of satelltes n tme s peroc wth the pero of approxmately 4 hours. Fg. 5 splays the hstogram of HPE. It shows that the hghest number of occurrences has a value of about 0.8 meters. Fg.6 shows the hstogram of HPL. From the graph t s event that the hghest number of occurrences has a value of about 8.3 meters. 190

9 Fgure5: Hstogram of HPE. Fgure6: Hstogram of HPL. 6 COCLUSIO The ntegrty mechansm of EGOS system was analyze n the artcle. The complex queston of GSS ntegrty n a ralway envronment was escrbe. Equatons for the computaton of a horzontal protecton level were shown. Real measure EGOS ata were presente at the en. Accorng to the real ata t seems that the level of HPL s too hgh at tmes. There are some unexpecte hgh jumps n HPL tme behavor, at tmes tens of meters. Therefore there are relatvely frequent occurrences of false alarms. For the future capablty of usng EGOS n safety-relate ralway applcatons, t s necessary to etermne a methoology for fnng the real falure rates of the EGOS system at the earth groun. It s also necessary to fn new methos of processng ata from the EGOS system wth accorance to strct ralway stanars. More etale analyss of real measure EGOS ata wll be one n future work by means of statstcal an tme seres analyss. 191

10 Especally the tme seres analyss of HPE an HPL may be funamental for fnng the real falure rates of the EGOS system on earth groun. REFERECES [1] RTCA DO-9D Mnmum operatonal performance stanars for GPS WAAS Arborne Equpment. RTCA, Inc., Washngton, D.C., 006. [] Flp, A. Integrta bezpečnost a spolehlvost systémů EGOS a Galleo [onlne]. 011 [ct ] Dostupné na: < [3] Flp, A.: Whch of EGOS avgaton Moes for Ralway Sgnallng: Precson Approach or En Route? Internatonal Symposum on Certfcaton of GSS Systems & Servces (CERGAL 010), Rostock, Germany, Aprl 8-9, 010. [4] Roturer, B., E. Chatre, J. Ventura-Traveset. The SBAS Integrty Concept Stanarse by ICAO. Applcaton to EGOS-ESA, EGOS for Professonals, Publcatons, GSS Conference, May 001 [5] Oehler.,V. Galleo Integrty Concept an performance. ESA ocument no. ESA-DEUI-G- T/01331,005. [6] Knesslan, F., Stubber, C. Combne ntegrty of GPS an Galleo. Workng Papers. January/February 010. InseGSS. [7] kforov, I. V., Choquette, F., & Belgum, A. T. (003). Integrty Equatons for Safe Tran Postonng Usng GSS. Insttuto Italno avgazone, Italan Insttut of avgaton, 171. The artcle was prouce uner the support of the project Transport R & D Centre ( CZ.1.05/.1.00/ ). 19