1 EFFECTS OF MASKING ANGLE AND MULTIPATH ON GALILEO PERFORMANCES IN DIFFERENT ENVIRONMENTS M. Malicorne*, M. Bousquet**, V. Calettes*** SUPAERO, 1 avenue Edouard Belin BP 43, 3155 Toulouse Cedex, France. C. Macabiau**** ENAC, 7 avenue Edouard Belin BP 455, 3155 Toulouse Cedex, France. Abstract Key words: Satellite navigation syste, Galileo, ultipath, urban environent, ray launching Over the past few years, any applications of satellite navigation systes have been developed. Aong the wide range of applications of such systes, transportation in urban environent sees to be one of the ost proinent. Hence it is iportant to fully and accurately characterize the receiver perforance for this application. The urban environent is characterized by high asking angles and the presence of a great nuber of obstacles which produce ultipath. Investigating the receiver perforance in this ediu requires a odel of the wave propagation. In this paper, a ray launching siulation tool is used to characterize different environents and the receiver errors due to ultipath and the receiver perforance with and without augentation are evaluated. Introduction Applications of satellites navigation syste are expending rapidly. Europe has initiated the Galileo progra, targeting the ipleentation of a navigation syste independent and interoperable with GPS and Glonass. The architecture of Galileo is based on MEO constellation of 3 satellites with 3 spares at an altitude of 3616 k, with an inclination of 56, distribute over 3 planes. Galileo ust take into account the need of a large nuber of applications. The requireents of an urban user ay be very stringent. We ay assue that an urban user needs availability greater than 9% with an accuracy better than 1 eters. However eeting the user needs generally is not an easy task since the urban ediu presents quite constraining conditions which do not always enable to copute the user location at the level of accuracy required. The first section of this paper deals with the channel odel. As the wavelength is sall with respect to the obstacles size, it is assued that the signal propagation can be odeled as a ray using the geoetric optic principle. Using such principle, siulation tool have been developed which coputes the intersections between the rays and the facets that constitute the environent. In a second part, we study the influence of ultipath on the receiver. Finally, eans to eet user perforance objective are proposed. Additional sensors are considered to iprove the receiver perforance. 1. Channel odel Multipath propagation is alost inevitable in ost satellite navigation ground applications, since there are any obstacles as buildings in the surroundings of the receiver. Several ethods exist to estiate the effect of these perturbations. Methods based on statistical forulations fro peculiar easureent have been disregarded since they cannot provide the required level of precision. These ethods do not provide a realistic odel of asking angle and ultipath. Moreover, these ethods are valid only for soe specific type of environent, range of frequencies, etc Therefore, a deterinistic ethod, based on geoetrical optic, has been selected that allows to describe the environent in 3D and ultipath propagation to the desired level of accuracy. It allows to siulate different transitters and receivers, fixed or obile. * Ph. D. candidate, e-ail : arie.alicorne@supaero.fr ** e-ail : ichel.bousquet@supaero.fr *** e-ail : vincent.calettes@supaero.fr **** e-ail : acabiau@recherche.enac.fr This work is supported by Alcatel Space Industries and is carried out in the fraework of TeSA laboratory. on Integrated Navigation Systes
This siulation tool is coposed of four odules. The first one defines the characteristics of the satellite constellation, several constellations can be siulated (Walker, GEO, ). The second odule siulate the wave propagation. This odule deterines for each satellite the direct path and the coordinates of the signal ipact on the environent surfaces. The next odule is the electroagnetic odule which deterines, for every path, the attenuation introduced by the reflection and the delay between the direct path and the reflected path. At each reflection, the reflection coefficient is coputed according to the nature of the surface. Moreover, a reflection translates in a polarization change. For each path, the siulation tool deterines the signal polarization, then the signal is decoposed in two signals polarized right hand circular and left hand circular. This allows to obtain the coposite signal at the output of the receiver antenna considering antenna gain on each polarization. The last odule defines the characterization of the scene. Different paraeters can be used for siulate the scene : Building surface Building height Street length The siulation tool can also used VRML files to define the scene. Figure 1: Siulation tool interface Figure 1 shows the siulation tool interface with a VRML scene. In red, it is the direct path and in blue the reflected path. The facets encountered are in blue.. Influence of ultipath on the receiver In a navigation receiver, there is an acquisition ode and a tracking ode. In this paper, it is assued that the receiver operates in the tracking ode. A non coherent delay lock loop is considered as the advantage of the non coherent loop with regard to the coherent loop is to avoid the need for carrier phase estiation. Multipath is characterized by the su of delayed echoes of the eitted signal, hence the received signal could be written as follows: r N s ( t) = a ( t) p( t ( t))sin(πf t + φ ( t)) = p Where N is the nuber of reflected paths, a the aplitude of path, τ delay of path and φ = πf dt πf pτ, f is the Doppler shift of path. In the following, the teporal dependence of the d paraeters are not taken into account. The influence of ultipath depends on their signal strength and delay copared to that of the line of sight signal. The delay loop induces a tracking error on direct path delay estiation called «code offset». on Integrated Navigation Systes
3 Figure : Non coherent delay lock loop phase detector In a non coherent DLL, the S curve is obtained by subtracting the squared early and late responses: S nc ( ˆ τ where : (τ ) = jφ e dt t M M jφ ) = ar( ˆ τ + d ) e ar( ˆ τ d ) t T = c R is the correlation function of the code (t) d is the early-late spacing d = Tc where T c is the code period ˆ τ is the estiate of the receiver line of sight delay p, correlated over a period of T p seconds. τ. The one-sided noise bandwidth 1 T c of the averaging operation is referred to as the tracking loop bandwidth B L. To understand the effects of ultipath on the code tracking, it is iportant to consider two situations: The fading bandwidth B F is larger than the tracking loop bandwidth B L : fast fading The fading bandwidth B F is saller than the tracking loop bandwidth B L : slow fading.1. Fast fading If B F is large copared with B L, all cross products are filtered out because of their relatively high frequencies, so the resulting S-curve is siply the suation of M+1 different non coherent DLL S curves. S nc M [ ( ar( ˆ τ + d ) ) ( ar( ˆ τ d ) ] ( ˆ τ = τ ) ) =.. Slow fading If B is sall copared with F ultipath tracking errors. B L, then the averaging over c T seconds has no influence on the resulting S nc M M jφ ) = a ( ˆ ˆ R τ + d ) e ar( τ d ) = = ( ˆ τ e jφ In the case of one reflected path (M=1), axiu delay errors occur if the reflected path has a phase difference of or pi radians with the line of sight path. This could observed in Figure 3. on Integrated Navigation Systes
4 Figure 3: Multipath tracking errors function of SMR Figure 4: Multipath tracking errors function of PhiR Figure 3 presents ultipath tracking errors functions of different paraeters, where SMR = a a (ultipath 1 aplitude to signal aplitude ratio), TauR = τ 1, PhiR = φ1 φ. Ones can note that the higher SMR, the larger the code offset is. On the other hand if the delay between the reflected path and the direct path is sall, the code offset is low. So, if considering outputs of Figure 3 and Figure 4, ones could expect that the ultipath tracking error will be not very large. The syste perforance in the different environents is now evaluated. 3. Perforance easureent for different environents 3.1. Measureent results These easureents were perfored in collaboration with ENAC (Ecole Nationale de l Aviation Civile). Four environents have been considered: Urban Industrial Residential High buildings with open area The receiver has been operated in a car with a roof agnetic antenna for about an hour in each type of environent. A laptop coputer recorded easureents every second, easureents being the nuber of GPS satellites in view. Figure 5 : Solution status in urban ediu Figure 6 : Solution status in industrial ediu on Integrated Navigation Systes
5 Figure 7 : Solution status in residential ediu Figure 8 : Solution status in ediu with high buildings Figures 5, 6, 7, 8 present the solution status during the easureents for the different environents. A zero value is valid solution, a one corresponds to a lack of observations and value equal to two or above corresponds to coputed solution but considered as not reliable. In Figure 5, for urban ediu, position evaluation is not often coputed because of a lack of observations, and at ties, the solution is not reliable because of a bad geoetry of the satellites in visibility. For industrial and residential environent, visibility of the satellite is good but reliable solution are not often obtained. In the environent coposed of high buildings with open areas, the satellites are less asked than in urban ediu. However when the receiver is near to a building, the satellites are asked only on a one side and as the satellite geoetry is bad, the solution is not reliable. In the next section, a coparison of visibility statistics obtained with easureent and with siulation is studied. 3.. Measureent and siulation coparison Visibility statistics obtained with the siulation tool are copared with easureents realized in four different environents. The following table shows the paraeters used to define the scene with the siulation tool for the different environents. BUILDING SURFACE BUILDING HEIGHT STREET LENGTH Width Length Mean Standard () () () () deviation () Urban 3 5 6-5 RESIDENTIAL 11 1 6 3 7 INDUSTRIAL 1 31 8 19 HIGH BUILDING 5 9 3 1 45-6 Table 1 : Scene paraeters on Integrated Navigation Systes
6 Figure 9: Coparison between easureents and siulation results The Figure 9 presents histogra against the nuber of satellite in visibility. It copares actual easureents perfored in four different environents (thin line) and the siulations realized with the ray launching with the paraeters gives in table 1 (bold line). In the following, all results are obtained with our siulation tool. Statistics on ultipath, considering the Galileo constellation, are now siulated in the four environents already defined. The siulation lasts five inutes, with a sapling period of one second. These statistics ake possible to copare the different environents and to characterize the different path which reach the receiver. In the siulations, we consider that a satellite is only visible if there is a direct path, as the receiver can not track a reflected path. Figure 1 : Path characteristics in urban ediu Figure 11: Path characteristics in industrial ediu on Integrated Navigation Systes
7 Figure 1: Path characteristics in residential ediu Figure 13: Path characteristics in ediu with high buildings The Figures 1, 11, 1, 13 shows the relation between the nuber of reflections, the gain and the delay of each reflected path for the different environents. It appears that the higher the nuber of reflections, the higher the delay and the attenuation on the path. It is interesting to note that in the residential environent (Figure 1), delays are shorter as the obstacles are closer fro the receiver than in the other case. Moreover, for three reflections, the gain is already less than.1, this confirs that it is useless to consider ore than three reflections. 4. Galileo perforance The carrier used in the siulations is in L band, near L1 GPS frequency, and is right hand circular polarized. The signal has a rectangular wave for with a chip rate of 1.3 MHz (it does not represent Galileo baseline but rather a standard case). The siulations is carried out in Toulouse, France (latitude:43.6, longitude:1.45 ) and the receiver is oving at a velocity of 3k/h. Pseudo-range easureents are available, ionospheric and tropospheric correction are perfored using odels in the receiver. The Kalan filtering applies on easureents. The state vector is coposed of the receiver position, the receiver speed and the receiver clock error. The propagation odel considers that user speed is constant. The syste perforance on the user position deterination is evaluated for five different cases. A coparison will be carried out with : Urban environent Industrial environent Residential environent Environent with high buildings with soe open areas The analysis focuses on the horizontal error, as vertical error has little iportance for urban user. Figure 14: Percentage of tie function of position error for different environents on Integrated Navigation Systes
8 It is assued that the perforance objectives for an urban user corresponds to an accuracy between 1 and 1 eters, with axiu availability (at least 9%). Figure 14 shows syste perforance without augentations. In the industrial environent and the environent with high building and soe open area, the results are satisfactory. In the case of urban environent and of residential case, the results are worst than in the two precedent cases. More generally the perforance is not sufficient in all situations. Since the stand alone syste cannot provide the level of accuracy desired, several type of augentation are to be considered to iprove the syste perforance. There are several approaches to iprove the syste perforance: Space based augentations : copleenting the Galileo constellation by the GPS constellation Add additional sensors at the user level : altieter, dead reckoning 5. Augentations at the space segent One approach to iprove the availability of the syste is to increase the nuber of satellites, for instance copleenting the Galileo constellation with GPS constellation. We ake the assuption that receiver can be used in dual ode (Galileo + GPS), with a known tie bias between GPS/Galileo. 1 3 Urban, Galileo % of tie, horizontal error < X 9 8 7 6 5 4 3 Urban, Galileo Urban, Galileo + GPS % of tie, nuber of satellites = X 1 4 6 8 1 1 14 16 18 15 1 Urban, Galileo + GPS 1 5 5 1 15 5 3 35 4 45 5 Horizontal error Figure 15: Percentage of tie function with and without GPS 4 6 8 1 1 14 16 18 Nuber of satellites in view Figure 16: Nuber of satellite in view Figure 15 copares the perforance of the Galileo syste alone and the perforance of Galileo copleented by the GPS constellation. An iproveent is observed but it is sall. Overall the required objectives are not et. Figure 16 shows the nuber of satellites in view when there is only Galileo constellation and when it is copleenting by GPS constellation. 6. Galileo perforance with additional sensors The second approach to iprove the syste perforance is to hybridize the receiver with additional sensors. Two types of sensors have been considered : Baro_altieter Dead reckoning (sensor providing data about distance and direction) odeled as a planietric sensor The considered sensor ipairents are a noise, a bias and a drift. The basic idea is to use sensors when there is a lack of easureents, and to use the navigation syste to recalibrate the sensors. The drawback of this solution is soe coplexity induced at the receiver architecture level. The state vector of the Kalan filtering is the sae than previously with addition of sensor bias. In addition of the pseudo-range, the Kalan filter uses the sensor readings. Now, the syste perforance with each of the two sensors will be analyzed. In the following siulations, only the case of the urban ediu is considered. on Integrated Navigation Systes
9 6.1. Altieter The first sensor considered is the baro-altieter. In the following siulation, the sensor noise is estiated to be.1 eter and its bias 1 eter. Different values of drift are considered. Figure 17 :Percentage of tie function of position error without and with altieter Figure 17 presents the percentage of tie function of horizontal error without augentation and with different type of baro-altieter. Thanks to additional easureent obtained fro the sensor significant iproveent of perforance are obtained, in particular for the 5k/h case because of the greater lack of easureents without augentations. Overall, the results are good but not sufficient with regard to the objectives. 6.. Dead reckoning The second sensor considered is dead reckoning which is odeled like a planietric sensor. Figure 18 : Percentage of tie function of position error without and with dead reckoning Figure 18 shows the percentage of tie function of position error without augentation and with dead reckoning for different drift and noise. Whatever sensor drift have been considered, in the both case, we obtained the desired result. Of course with a sensor of best quality, the result are better, but a sensor cheaper is sufficient to fulfil the user needs while taking into account two ajor drivers of this study: user cost and syste cost. on Integrated Navigation Systes
1 Conclusions There is an increasing requireent for an accurate navigation syste for urban application. Urban user needs an accuracy between 1 and 1 eters, with an availability at least 9%. The presence of ultipath coponents in the received signal adversely affects the ability of the receiver to copute accurate navigational solutions. The perforance in stand-alone of the Galileo syste in urban environent is not sufficient because of these two phenoenon, so different ways to iprove perforance have been considered. The first ethod which consist in copleenting Galileo constellation with GPS constellation, iproves result but not sufficient with regard to our objectives. The second ethod is the hybridization with additional sensors yields very proising results. Acknowledgeents Acknowledgents are due to Mr. Bruno Lobert and Mr. Christophe Bourga fro Alcatel Space Industries for their support during this study. References 1. R.D.J. Van Nee, Spread Spectru code and carrier synchronization errors caused by ultipath and interference, IEEE transactions on aerospace and electronic systes, October 1993, v9,n 4, pp. 1359-1365.. M. Malicorne, M. Bousquet, C. Bourga, B. lobert, Ph. Erhard, Galileo Perforance in urban environent, ION GPS, San Diego, June. 3. M. Malicorne, M. Bousquet, V. Calettes, Influence of ultipath on GNSS receiver perforance in urban environent, 19 th AIAA international counications satellite systes conference, Toulouse, 17- April 1. 3. Parsons, J.D., The Mobile Radio Propagation Channel, Pentech Press, London, 199 4. www.ergos-fr.co on Integrated Navigation Systes