Mayjune07 064-073.Pdf
working papers Envisioning a Future GNSS System of Systems Günter W. Hein, markus irsigler, Jose Angel Avila-Rodriguez, Part 3 Stefan Wallner, Thomas Pany, Bernd Eissfeller, and Philipp Hartl A Role for C-Band?
MEASAT-3 satellite with C-band and K-band transponders, The Boeing Company photo The radio frequency spectrum is finite. Crowding in the L-band occupied — or planned for use — by the world’s global navigation satellite systems is only going to get worse, as more systems and more signals come on line. Several gigahertz up the RF spectrum from L-band, however, lies a wide swath of bandwidth that is comparatively untapped: the C-band. In fact, early in its development the Galileo program received an allocation of C-band from the World Radiocommunications Conference. Although Galileo system designers decided quite early not to use the allocation for a variety of practical reasons, C-band remains an enticing subject because of certain characteristics that seem to complement or compensate for technical limitations of L- band, particularly the need for better indoor positioning capability. This column examines C-band as a candidate for a future GNSS signal or signals and evaluates its advantages and disadvantages compared with L-band signals.
he radionavigation satellite portion of the C-band had been assigned tions. We will refer to these aspects in service (RNSS) portion of the for radionavigation, but technical com- detail in the following discussion. RF spectrum is overcrowded, plexities made it impossible for the first As happens with any kind of tech- T especially on L1 where GPS, generation of Galileo to make use of it. nology, however, many ideas that have Galileo, Compass overlap portions of Phase noise problems, increased signal been abandoned in the past due to exces- one another’s signal frequencies and transmit power requirements, and signal sive technical challenges or demanding GLONASS signals occupy more than attenuation issues — to name only a few drawbacks often become objects of inter- 11 MHz of nearby bandwidth. Indeed, — knocked down all the proposed solu- est some years or decades later. As tech- even those bands that have not been used so far will certainly be shared by many systems in the near future. There- fore, the search of alternative frequency resources is something that must inevi- tably occur with a high probability in the coming years. During the World Radio Conference EHF Extremely high frequency (Microwaves) VHF Very high frequency SHF Super high frequency (Microwaves) HF High frequency 2000 (WRC-2000), the Galileo program UHF Ultra high frequency MF Medium frequency obtained authorization to use C-band FIGURE 1 Radio Spectrum [Source: Industry Canada] frequencies. At the time, a dedicated
64 InsideGNSS may/june 2007 www.insidegnss.com working papers
nology evolves, constraints alter, and the between 3.5 GHz and 8.0 GHz, the study, see the article by M. Irsigler list- environment of possibilities changes. definition of C-band for satellite com- ed in the Additional Resources section Against this background, the ques- munications ranges from 3.5 GHz to 6.4 near the end of this column.) The filed tion emerges as to whether the use of C- GHz. C-band portion of Galileo is partitioned band frequencies could represent a real Specifically, C-band frequencies are into the uplink service band (5000 - 5010 alternative for a future GNSS. In this col- preferred for geostationary satellites, MHz) and the RNSS band between 5010 umn, we will try to shed some light on generally occupying the uplink fre- MHz and 5030 MHz. this interesting possibility. Before that, quency band from Band TX Frequency RX Frequency let us first look at what we understand 3.6 GHz to 4.2 GHz about C-band and how the regulatory and downlink fre- Extended C-Band 5.850 - 6.425 GHz 3.625 - 4.200 GHz RF spectrum situation affecting its use quencies between Super Extended C-Band 5.850 - 6.725 GHz 3.400 - 4.200 GHz varies in different countries. 5.8 GHz to 6.4 GHz. INSAT C-Band 6.725 - 7.025 GHz 4.500 - 4.800 GHz Moreover, in optical Palapa C-Band 6.425 - 6.725 GHz 3.400 - 3.700 GHz C-Band Definition networks the con- Russian C-Band 5.975 - 6.475 GHz 3.650 - 4.150 GHz The general definition of C-band refers ventional C-band is to the portion of the electromagnetic defined in terms of LMI C-Band 5.725 - 6.025 GHz 3.700 - 4.000 GHz spectrum in the microwave range of wavelengths rang- TABLE 1. C-band frequency variants at the satellite frequencies between 4 and 8 GHz. It ing between typical was the original frequency allocation values of 1,530 and 1,560 nanometers. Thus, a future C-band signal could for communication satellites, including In addition to these application- use the frequency band between 5010 those primarily in use today. based categories, slight variations of C- MHz and 5030 MHz, offering a rather Typical antenna sizes on C-band– band frequencies are approved for use limited bandwidth of 20 MHz but allo- capable user equipment range from in various parts of the world. Table 1 cated in a frequency band not yet over- 2.5 to 3.5 meters on consumer satellite presents some of these. loaded by other systems. Although 20 dishes, although larger ones and smaller As an example of the diversity in MHz does not seem to be a very broad ones can also be used depending upon the definition of the C-band in different bandwidth (especially if we compare it signal strength. As one can imagine, countries, Figure 2 and Figure 3 show the with the well known L-band), one of the smaller antennas are of special interest particular regulatory situations in the initial drivers to use the C-band was that for radionavigation purposes. United States and Canada, respectively, it could solve the limited frequency band Let’s take a closer look at where C- for the frequency range from 3 GHz resources of GNSS. band lies on the radio spectrum. Figure to 7 GHz. As we can see, each country The exact carrier frequency of a C- 1 shows the current radio bands as cat- presents slightly different allocations for band signal would be 5,019.86 MHz, egorized in wavelength and frequency the different services. resulting in a carrier wavelength of only domains. As we can clearly recognize, 6 centimeters. The use of such a short the C-band falls between the ultra high Could C-Band Work for carrier wave has a significant effect on frequency (UHF) and super high fre- Galileo? various aspects of signal propagation quency (SHF). In comparison, the L- Although not considered for the first and signal processing and comprises a band corresponds to UHF. generation of European Galileo satel- series of benefits and drawbacks that we On closer inspection, some degree lites, the (additional) use of C-band fre- will address next. of arbitrariness occurs in defining the quencies was subject of a study carried Characteristics of the C-Band. The boundaries of the C-band, depending out back in the year 2001 shortly after application of C-band navigation signals on the technical world we are moving WRC-2000 delegates assigned the fre- provides both advantages and draw- in. Although in microwave techniques quency band between 5000 MHz and backs. At C-band, the increased free C-band refers to the frequency band 5030 MHz to Galileo. (For details of this space loss represents the most significant issue. Despite its name, “free space loss” actually refers to the fact that an omni- directional antenna must not exceed a certain dimension, a constraint that relates to the signal wavelength. Thus, an omnidirectional C-band antenna (for 5 GHz) is 3.2 times smaller LF Low Frequency ELF Extremely low frequency in linear dimension than an L-band VLF Very Low Frequency VF Voice frequency antenna (with a 19-centimeter wave- length at 1.575 GHz). Its area is 10 times FIGURE 1 Radio Spectrum [Source: Industry Canada] smaller than that of a standard L-band
www.insidegnss.com may/june 2007 InsideGNSS 65 working papers
(which would pick up more energy from the satellites), but those antennas inevitably have a directional characteristic. Over- coming this major issue would require significant modifi- cations with respect to the satellite pay- load and/or the user equipment. R e s e a rc h h a s identified other sig- nificant drawbacks to C-band includ- ing increased sig- nal attenuation due to foliage or heavy rain as well as nega- tive effect on signal tracking. On the other hand, C-band’s smaller ionospheric errors and technolog- ical progress might balance some of the FIGURE 2 United States Frequency Allocations within the C-Band (October 2003) [US Department of Commerce. National disadvantages from Telecommunications and Information Administration. Office of spectrum management] a long-term point of view. C-Band’s Suitability for GNSS In order to assess the feasibility of future C-band technology for satellite naviga- tion purposes, we will recapitulate the major findings of Irsigler’s 2000-2001 study for DLR men- tioned earlier, which examined various aspects of signal FIGURE 3 Canada Frequency Allocations in the C-Band. This chart is based on the 2000 Canadian Table of Frequency Alloca- propagation and sat- tions, which bases on decisions of World Radio Conferences, including WRC-1997. For further information on spectrum or radio matters, contact the Spectrum and Radio Policy Directorate, Industry Canada, Ottawa. ellite signal tracking at C-band. To allow antenna. As a result, a C-band antenna independent of the frequency because comparison between the current GPS picks up only one-tenth as much broad- the loss is only proportional to the and the planned Galileo system, we cast signal power as L-band. inverse of the squared distance between compare the performance expected at C- The free-space-propagation-loss in transmitter and receiver. One can, of band to that of the L-band under similar the antenna’s far field region is actually course, build bigger C-band antennas or identical conditions.
66 InsideGNSS may/june 2007 www.insidegnss.com working papers
Signal Parameters GPS L1-C/A Assumed Galileo C Band Parameter C L Factor f [MHz] 1575.42 5019.86 Free space loss - + -10.0 dB Carrier Wave λ [m] 0.19 0.06 Ionospheric path delay + - +10.0 Chipping Rate [Mcps] 1.023 8.184 Ionospheric amplitude scintillation + - +5.6 Chip Length [m] 293.05 36.63 Ionospheric phase scintillation + - +3.1 Data Rate [bps] 50 150 Ionospheric refraction + - +10.0 Predet. Int. Time [s] 0.02 0.0067 Ionospheric Doppler shift + - +3.0 Bandwidth [MHz] 2.046 20 Tropospheric path delay o o --- Chip Shape/Modulation Scheme BPSK RC Tropospheric amplitude scintillation - + -2.0 TABLE 2. GPS and Galileo C-Band signal parameters Tropospheric phase scintillation - + -3.0 Attenuation by water vapor and oxygen (worst case) - + -0.2 dB Table 2 lists the relevant Galileo and GPS signal parame- Rainfall attenuation (worst case) - + -4.5 dB ters. Unless otherwise stated, this column bases all subsequent Attenuation by clouds and fog (worst case) - + -0.8 dB computations, diagrams, and tables on these parameters. Foliage attenuation - + -1.0 dB Interestingly, instead of using BPSK or BOC modulation, a TABLE 3. Signal propagation at L- and a C-Band raised cosine (RC) pulse shaping scheme with a chipping rate of 8 Mcps was then assumed in the study to be implemented for a C-band Galileo signal. The shape of a raised cosine chip Link Budget Parameter C-Band L-Band is defined by the so-called roll-off factor. For the Galileo C- Received Power Level -163 dBW -163 dBW band signal, a roll-off factor of 0.22 was assumed. Total Sign. Attenuation 204.8 dB 189.3 dB RC-signals are band-limited signals and were abandoned Gain Satellite Antenna 14 dB 14 dB for the L-band very early in the Galileo program due to the fact that the resulting services would have very limited assigned Gain User Antenna 0 dB 0 dB bandwidths with a handicapped performance from the very Required Satellite 27.8 dBW 12.3 dBW beginning. In fact, no matter how much we would increase the Antenna Input Power 602.6 W 17.0 W receiver bandwidth, given the limited bandwidth of the trans- TABLE 4. Required minimum satellite antenna input power (C-Band vs L-Band mitted signal, no improvement in the positioning performance payload). Total signal attenuation includes free space loss, worst-case tropo- could be obtained. spheric attenuation, polarization mismatch loss and antenna depointing losses, see Table 5. Although this was true in the L-band where many services had to co-exist, in the C-band wider bandwidths are expected and, thus, if the number of services is reduced, the use of this C-band signal is attenuated up to 4.5 dB greater than an L-band modulation scheme could potentially bring some benefits, signal. Foliage attenuation, which is assumed to be around 1 especially regarding the limitations of the emissions. dB/m at L-band and around 2 dB/m, can also be an issue. Although this was true in the L-band where many services Effect of C-Band on Power Budget. The main signal propaga- had to coexist, in the C-band wider bandwidths are expected. tion parameter that affects a satellite payload design is signal Thus, if the number of services is reduced, the use of this modu- attenuation. Compared to the L-band, free space loss and rain lation scheme could potentially bring some benefits, especially attenuation are significantly greater at C-band. regarding the limitations of the emissions. In order to compensate for the increased signal attenuation, Signal Propagation at C-Band. Table 3 compares the various a C-band signal will have to be much stronger (increased trans- signal propagation characteristics of L- and C-band. Benefits mit power) than an equivalent L-band signal. Otherwise, if we with respect to the other frequency band are indicated by “+” assume identical satellite transmit power at L- and at C-band, whereas drawbacks are indicated by “-”. The quantitative dif- the received C-band signal will be much weaker. ference with respect to the L-band is assessed in the last col- The following reverse computation of the minimum transmit umn. power assumes that the power level of a future C-band signal for As can be derived from Table 3, all ionospheric effects are a 0 dBic antenna is –163 dBW. Assuming that the noise density is Table less severe at C-band. The expected ionospheric path delays are N0=-204 dBW/Hz, the corresponding C/N0 is 41 dB-Hz. smaller than at L-band by a factor of 10, and scintillation effects 4 summarizes the results of these calculations. become less significant as well. Normally, the specified received power level needs A major issue of using C-band frequencies is the increased to be much higher than -163 dBW to provide a good
influence of signal attenuation. Due to the increased free space (C/N0)eff within the tracking loops. The following computation loss at higher frequencies, a C-band signal transmitted at iden- is based on the requirement that the signal should be tracked
tical transmit power as an L-band signal is 10 dB weaker when with a C/N0 of at least 45 dB-Hz, a value that is easily obtained it arrives at the user antenna. Moreover, in case of heavy rain, a for GPS signals. Table 5 compares the computation of the
www.insidegnss.com may/june 2007 InsideGNSS 67 working papers
Link Budget Parameter Unit C-Band GPS L1 (assuming identical antenna is also a suitable approach to
Effect. C/N0 (tracking loop) dB-Hz 45 45 conditions at both cancel out multipath and/or interfer- Implementation loss dB 6 6 bands). ing/jamming signals. C/N @ user antenna output dB-Hz 51 51 • Minimization of Receiver Implemen- 0 Higher C- tation Losses. As we derived in Table Power level (user ant. output) dBW -153 -153 Band Signal 5, the effective C/N0 also depends Gain of user antenna dBic 3 3 Power on the receiver implementation loss. Power level (user ant. input) dBW -156 -156 Due to the increased Because the C/N0 of the received Depointing loss (user) dB 0.25 0.25 free space loss and C-band signal will be much lower Polarization Mismatch Loss dB 3 3 tropospheric atten- than an equivalent L-band signal uation, a future C- (assuming identical conditions), no Tropospheric attenuation dB 5.9 0.4 band signal will be additional losses can be accepted. Free space loss (E=10°) dB 195.4 185.4 approximately 10- In Table 5, an implementation loss Depointing loss (satellite) dB 0.25 0.25 16 dB weaker than of 6 dB has been assumed. This is EIRP dBW 48.8 33.3 an L-band signal a typical value for low-end receiv- Gain of satellite antenna dBic 14.0 14.0 when received at ers. In high-quality receivers we dBW 34.8 19.3 the ground (assum- can expect implementation losses Required satellite antenna input power ing identical satellite of only 1-2 dB. Especially at C- W 3020.0 85.1 transmit power and band, the implementation losses TABLE 5 . Required satellite antenna input power to provide a C/N0 of 45 dB-Hz identical user anten- should be as small as possible. The within the receiver tracking loops. For the tropospheric attenuation at C-band nas). Signal strength main parameters that determine the the major contributions are a worst-case rainfall attenuation of 4.6 dB and attenuation by clouds and fog of 0.9 dB. The corresponding L-band values are at the user antenna receiver implementation loss are the much smaller (0.1 dB, respectively). also determines the LNA (low noise amplifier) noise fig-
(C/N0)eff with which ure and the quantization process of required satellite antenna input power the signal is tracked within the tracking the A/D (analog/digital) conversion. for C-band and GPS L1. The results loops. With respect to the A/D conver- are obtained considering the following We consider two general measures to sion, the quantization process causes parameters: compensate for the 10–ß16 dB loss at C- signal degradation, which depends • Receiver Implementation Loss: 6 dB band: Increasing satellite transmit power on the resolution of the quantiza- (low-end receivers) in conjunction with implementing suit- tion process. It can be shown that the • Maximum atmospheric attenuation: able antenna/receiver design. As energy actual signal degradation decreases see Table 5 is a limited resource on board a satellite, with increasing resolution. A mini- • User antenna gain: 3 dB the following section will concentrate mum of two bits is required to limit • Satellite antenna gain: 14 dB more on antenna design as a means to the signal degradation to 1-1.5 dB.
• Noise density: -204.0 dBW/Hz guarantee sufficiently high (C/N0)eff at The use of 3-5 bit quantization can
To provide an effective C/N0 of 45 C-band. reduce the corresponding signal dB-Hz, the satellite antenna input power The following approaches are suit- degradation down to 0.5-0.7 dB, as at C-band will have to be approximately able to achieve this goal: described in the article by B. Parkin- 35 times higher than at L-band. We must • Use of Phased Array Antennas. In son and J. Spilker cited in Addition- note that the preceding calculation of the contrast to standard omnidirec- al Resources. As a result, multi-bit tropospheric attenuation includes rain- tional user antennas, phased array quantization is strongly recommend- fall attenuation and attenuation due to antennas consist of multiple anten- ed for future C-band receivers. clouds, fog, water vapor, and oxygen. We na elements that are arranged in • Bite the bullet. Current high-sen- also need to emphasize that these val- the form of an array. By means sitivity receivers are known to be ues are the result of a calculation under of digital beam forming, several capable to cope with signal attenua- worst-case assumptions. antenna beams can be generated. tions of even more than 20 dB. Thus The actual required satellite antenna The beam widths and the result- a straight extension of this high input power strongly depends on the ing antenna gains depend on the sensitivity L-band signal processing receiver quality (implementation loss), number of antenna elements: With technology (mostly based on multi- the type of user antenna (phased array such an antenna design, typical ple correlators and optimized signal versus omnidirectional), and the actual gains of approximately 10 dBic can tracking/acquisition algorithms) is atmospheric attenuation. Whatever be achieved. Compared to a 3 dBic capable of tracking C-band signals. scenario is assumed, the required satel- standard omnidirectional user The accuracy is, however, reduced, lite antenna input power at C-band will antenna, the received power level can and such receiver designs ultimately be significantly higher than at L-band be increased by 7 dB. A phased array cannot compensate for any further
68 InsideGNSS may/june 2007 www.insidegnss.com working papers
- Loop Noise Bandwidth BL power reduction of C-band signals • Feasibility of Thermal Noise - Predetection Int. Time T σT by a “real” indoor environment. low-end C-band - C/N After having shown possible approaches receiver manu- 0 - Loop Order for loss reduction by improved antenna facturing using Oscillator Phase Noise - Loop Noise Bandwidth B designs at the user side, we are going to simple one-bit- (frequency instabilities) L ponder whether the power increase or qu a nt i z at ion - Clock Parameters h0,h-1,h-2 (σA,Rec , σA,Sat ) the sophisticated antenna approach is a techniques - Carrier Frequency f more promising path to follow. On the other - Loop Order side, an increase of - Loop Noise Bandwidth BL Transmit Power versus the satellite trans- Vibration Induced Phase Noise Phased Array Antenna mit power results in - G-Sensitivity of Oscillator σvib The use of phased array user anten- additional problems, - PSD of Vibration nas and the construction of high-end such as an increased - Carrier Frequency f C-band receivers with very low imple- power consump- - Loop Order mentation losses may not be necessary tion at the satellite, - Loop Noise Bandwidth BL if the satellite antenna input power can which would result Dynamic Stress Error e(t) be increased significantly. in additional or - Signal Dynamics (LOS) Increase of Satellite Transmit Power. larger solar panels. - Carrier Frequency f Many drawbacks of C-band navigation This, in turn, nega- TABLE 7. Signal and loop parameters affecting the PLL error sources. A more could be compensated by increasing the tively affects satel- detailed analysis can be found in [Irsigler, Eissfeller, 2002] satellite transmit power by a minimum lite launches (more of 10 dB. By means of this approach, space required within the satellite- Signal Tracking at C-Band the following enhancements can be launching rocket, increased satellite Table 6 summarizes the signal tracking achieved: weight, and increased launch cost). performance at L- and at C-band. Ben- • compensation of the increased free Use of Phased Array Antennas. At first efits with respect to the other frequency space loss sight, the use of phased array antennas band are again indicated by “+” while
• increase of (C/N0)eff within the sig- seems like a suitable approach to limit drawbacks are indicated by “-”. Note nal tracking loops, thereby reducing the required satellite transmit power that the classification in Table 6 is only the influence of thermal noise and and to compensate the occurring signal valid in cases that assume identical con- the probability of cycle slips while losses at C-band. The main advantages ditions in both bands (identical signal
enhancing the performance of the of this approach are the increased anten- structure, C/N0, smoothing constants, various lock loops (delay, frequency, na gain compared to an omnidirectional and so forth). phase) antenna and the ability to null out mul- Performance parameters related • compensation of the increased tropo- tipath and/or interfering/jamming sig- to code tracking (DLL performance, spheric attenuation, thereby increas- nals by means of beam forming. How- code noise, and code multipath) do not ing availability ever, the main drawbacks of using such depend on the carrier frequency; so, the • use of omnidirectional user anten- antennas are that they would be larger, two frequency bands perform similarly nas, resulting in a relatively simple heavier, more unwieldy, and complex in this regard. Due to the smaller carrier receiver architecture, moderate compared to omnidirectional antennas wavelength at C-band, phase noise, and power consumption, low manufac- and, consequently, their manufacturing carrier multipath are smaller than at L- turing costs, and enhanced mass- cost would be higher. Moreover, due to band (if expressed in meters). market suitability their increased size, they cannot be used Another benefit of C-band is the for certain applications. increased carrier smoothing efficiency. Parameter C L Moreover, because a phased array Due to the fact that multipath varia- Code tracking performance o o antenna consists of several antenna ele- tions occur more often at C-band, ments, a corresponding amount of front multipath effects can be smoothed out Phase tracking performance - + ends would be necessary (one front-end easily (assuming identical smoothing Code noise o o per antenna element). Additionally, a constants). However, major problems Phase noise + - beam-forming and beam-steering unit appear when we take a closer look at Code multipath o o would have to be implemented. In con- the carrier tracking performance at C- Carrier multipath + - trast to an omnidirectional receiver, band. Enhancing Poor PLL Performance. Carrier smoothing efficiency + - then, the phased array approach thus One results in complex receiver architecture, major drawback of using the C-band for TABLE 6. Signal tracking performance at L- and a thereby increasing size, weight, power satellite navigation is the very poor carri- C-Band. consumption, and manufacturing cost. er tracking robustness, as noted in Table
www.insidegnss.com may/june 2007 InsideGNSS 69 working papers
80 80 )[°] )[°] σ σ 60 60
40 40
20 20 15 15 tal PLL Jitter for C (1
otal PLL Jitter for L1 (1 0 10 0 10 To T 20 25 20 25 30 5 Acceleration [m/s2] 30 5 Acceleration [m/s2] 35 40 35 40 45 50 0 45 50 0 C/N0 [dB-Hz] C/N0 [dB-Hz]
FIGURE 4 Total PLL jitter for the GPS L1 signal (left diagram) and a C-band signal (right diagram) as a function of C/N0 and the signal dynamics (Line-of- sight (LOS) acceleration).
6. The dominant error sources of a phase analysis. The total PLL tracking error is oscillators (TCXOs) due to the more lock loop are the thermal phase noise σT, plotted as a function of the C/N0 and the stringent performance require- the oscillator phase noise induced by fre- line-of-sight acceleration. Additionally, ments. quency instabilities of the receiver and/ the tracking threshold defined by equa- • Use of high stable reference oscillators. or satellite clock (σA,Rec,σA,Sat), the vibra- tion (1) is plotted as a gray plane. The The oscillator’s frequency instability tion induced oscillator phase noise σvib, PLL can be deemed to be stable if the can be described by its Allan devia- and the dynamic stress error e(t). The total tracking error is less than the track- tion. As is the case for random vibra- occurring error sources depend on the ing threshold, i.e., for all parts of the sur- tion, frequency instabilities also result parameters listed in Table 7, and the PLL face that lie below the gray plane. in phase jitter. In order to reduce the can be deemed to be stable if the follow- The poor PLL tracking performance resulting phase errors, the use of a ing equation holds: at C-band is obvious. In this example, highly stable reference oscillator (for the total PLL jitter is mostly higher example, an oven controlled crystal than the tracking threshold defined oscillator, OCXO) instead of a stan- by equation (1). Even in cases of only dard TCXO should be considered. Oscillator phase noise, vibration weak accelerations, the receiver would The main drawbacks of this induced phase noise, and dynamic stress not be able to track the signal, and if approach are increased cost and error are proportional to the carrier fre- several loop, clock, or vibration param- power consumption compared to a quency. As a result, we can expect a sig- eters are modified, the PLL performance standard TCXO. (Note that in mass- nificant increase of PLL jitter at C-band. at C-band is always much poorer than market GPS receivers, even lower Thus, we would expect the PLL perfor- at L-band. quality quartz oscillators are in use). mance at C-band to be much poorer In order to improve the PLL per- Additionally, the influence of fre- than at L-band. This statement can be formance at C-band, the following quency instabilities also depends on verified by means of a PLL performance approaches can be taken into account: the loop noise bandwidth. A reduc- analysis which is based on the following • Enhancement of the reference oscil- tion of the resulting phase jitter can assumptions: lator’s g-sensitivity. The influence of principally be achieved by increasing • Non-coherent carrier tracking (e.g. random vibration strongly depends this bandwidth. However, the result- Costas Tracking) on the oscillator’s g-sensitivity. The ing enhancements are marginal and • Second-order PLL resulting phase jitter is directly pro- by increasing the loop noise band- • Loop noise bandwidth: BL=15Hz portional to the oscillator’s g-sen- width, the thermal noise correspond- • Predetection integration times: 0.02s sitivity so that a reduction of this ingly increases. (L-band), 0.0067 (C-band) parameter results in less phase jitter. • Significant increase of the loop noise • Carrier frequencies: 1475.42MHz (L- Random vibration is an issue espe- bandwidth. The dynamic stress band), 5019.86 MHz (C-band) cially for kinematic applications, error strongly depends on the loop • Satellite clock: Rubidium whereas such influences are not pres- noise bandwidth. The smaller the • Receiver oscillator: TCXO ent in static applications. The main loop noise bandwidth, the harder it • PSD of vibration: 0.05 g²/Hz drawback of this approach is that is to track the signal dynamics. On • Frequency of vibrations: appropriately optimized oscillators the other hand, increase of the loop
25Hz < fvib< 2500Hz are more expensive than standard noise bandwidth reduces the influ- Figure 4 illustrates the result of this temperature compensated crystal ence of dynamic stress and is, there-
70 InsideGNSS may/june 2007 www.insidegnss.com working papers
BENEFITS Ionospheric Path Delay Ionospheric Refraction All ionospheric effects are inversely proportional to the carrier frequency f (or a power Conclusions about C-Band of f). The ionospheric path delay at C-band, for example, is smaller than at L1 by a Performance Ionospheric Doppler shift factor of 10. As we have seen, C-band navigation Ionospheric Scintillation offers both benefits and drawbacks. By combining pseudorange observations at different carrier frequencies, the iono- Although it might be feasible to over- Iono-Free Linear spheric effect can be eliminated. The resulting observable, however, is significantly come the technical issues, it is still Combination L/C noisier than the individual pseudoranges. Compared to the L1/L2 observable, however, uncertain whether a (future) C-band an L1/C combination would be less noisy (by a factor 2.8). navigation system could ever compete The maximum carrier multipath error is a function of the carrier wavelength. Maximum Carrier Multipath with current sophisticated L-band errors at C-band would be 1/3 of the expected error at L1. equipment. However, a future C-band Carrier smoothing can be used to mitigate code noise and code multipath. Noise signal might be an interesting option in mitigation: The carrier smoothing process performs better at C- than at L-Band, as the ratio between carrier frequency and code rate is larger. Due to the small ionospheric combination with L-band signals. Carrier Smoothing influences at C-band, longer smoothing constants than at L-Band can be used. As discussed, many of the draw- Performance Multipath mitigation: Fading frequencies at C-band are larger than at L-band by a backs could be balanced by increasing factor of 3, i.e., multipath variations can be smoothed out more efficiently (assuming the satellite transmit power. This mea- identical smoothing constants in both bands). sure could also enhance the poor carrier Phase Tracking Accuracy The influence of thermal noise is proportional to the carrier wavelength. i.e., thermal tracking performance by reducing the Doppler Accuracy noise at C-band is by a factor of 3 smaller than at L-band. influence of thermal noise as well as the C-band antennas can be built three times smaller than L-band antennas. For miniatur- cycle-slip probability. It also leads to an Antenna Size ized applications, this could be an advantage. enhancement with respect to the avail- DRAWBACKS ability of the navigation service because it compensates for the increased signal Proportional to the squared carrier frequency, resulting in a range spreading loss at C- Free-Space Loss band which is larger than in L-Band by a factor of 10 (can be compensated by increase attenuation values at C-band. of satellite transmit power). Nonetheless, this measure would Depends on transmit power, antenna gains, implementation losses, and actual atmo- have negative effects on important pay- spheric attenuation. Assuming identical transmit power, antenna gains, and imple- load characteristics such as power con- Effective C/N0 mentation losses in both bands, the effective C/N0 at C-Band is 10-16 dB-Hz smaller sumption, weight, and size that, in turn, than at L-band (due to increased range spreading loss and atmospheric attenuation) increase the costs for manufacturing and Includes attenuation due to water vapor and oxygen (up to 0.2dB larger at C-Band), launching the satellites. Moreover, even Tropospheric Attenuation rainfall attenuation (up to 4.5 dB larger) and attenuation due to clouds and fog (up to if satellite transmit power is increased, 0.8 dB larger). Rainfall attenuation up to 4.6 dB must be expected at C-band. some basic problems such as the poor Typical values at L- and C-band are 1 dB/m and 2 dB/m, respectively. These values Foliage Attenuation carrier tracking performance would may strongly vary subject to foliage types and densities. still remain. In order to compensate for all the attenuation effects listed above, the satellite trans- Power Consumption We summarize the benefits and mit power would have to be increased significantly (with negative effects on satellite Payload drawbacks in Table 8. size, weight and launch costs). Includes amplitude scintillation, which is larger at C-band by a factor of 2, and phase Tropospheric Scintillation Potential Users of C-Band scintillation, which is larger by a factor of 3. As we mentioned earlier, the C-band Due to higher maximum Doppler shifts at C-Band (~11 kHz vs. ~6 kHz at L-Band), was not considered for the first genera- Signal Acquisition the Doppler search region increases. Assuming identical code lengths at both bands, tion of Galileo because of the technical signal acquisition takes about 1.8 times longer than at L-Band. implications that we outlined in this col- Carrier tracking robustness at C-band is much poorer than at L-band. It depends on the umn. Nonetheless, the European Space Carrier Tracking influences of thermal noise, oscillator phase noise, vibration-induced phase noise, Robustness and dynamic stress. Except for thermal noise, all influences are proportional to the Agency is planning a series of activities carrier frequency and, thus, larger than at L-band (by a factor of 3). in recognition of the fact that the C-band Depends on effective C/N , data rate, and oscillator phase noise. Due to increased could conceivably be used for navigation 0 purposes in the next decade. Cycle-Slip Probability oscillator phase noise (see “carrier tracking robustness”) and possibly smaller C/N0, the cycle-slip probability at C-band is expected to be larger than at L-band. Of course, C-band is not an exclu- The poor carrier tracking robustness can be enhanced by minimizing the oscillator’s sive asset of Galileo. Indeed, other global Oscillator Requirements g-sensitivity and by enhancing its frequency stability. This requires the use of high- navigation satellite systems have filed quality oscillators. for this band, although at the moment TABLE 8. Benefits and drawbacks for C-Band satellite navigation no one has used it yet for navigation. Moreover, also regional systems such as fore, another possible approach to that by increasing the loop noise the Indian Regional Navigation Satel- enhance the PLL performance. The bandwidth, thermal noise also lite System (IRNSS) and Japan’s Quasi- main drawback of this approach is increases. Zenith Satellite System (QZSS) have also
www.insidegnss.com may/june 2007 InsideGNSS 71 working papers
declared their intention of making use for Open Services. But can we imagine son for pulsing in mixed GNSS/pseudo- of the band. the opposite situation? lite applications is mainly to avoid the As we have seen in our analysis In addition, separating military and jamming of signals coming from space here, the use of the C-band raises seri- civil applications would also have other due to the higher power of terrestrial ous technical challenges, although, as consequences of interest. As we saw in pseudolite transmitters. The solution we have also pointed out, an increase of the first part of this series (January-Feb- to this near/far (strong/weak) problem power in the transmission of the satel- ruary 2007 issue), such an architecture was to transmit the total power only in lites together with directional antennas would allow further separation of the pulses. Although these individual pulses could solve the current drawbacks of the military and civil payloads, which might are very strong, the averaged pseudolite C-band. Moreover, once we solve these well accommodate the fact that many signal (taking into consideration the technical problems, the accuracy poten- civil and military applications work with absence of power in the interval between tial of the C-band could be used to help different standards. pulses) ultimately matches the level of solve many GNSS-related problems that Today civil and military payloads the weak signals from the satellite. we have today. are mounted together aboard the satel- Could we not perhaps apply similar The question that arises is: who lites so that in the final consideration ideas to C-band in order to avoid the could afford such directional antennas the more demanding standards of the high power figures that are needed to and such an increase of power? A first military have to be applied also to civil compensate for the higher propagation reaction would be to say that military components that might not require them losses of the C-band? Another approach or protected governmental applications at all. Moreover, consideration could might be to design C-band satellites that could fully satisfy such requirements be given to allow the civil community only serve users in certain locations even today. Can you imagine having access to the ground segment, which is and then allow satellite transmissions all the military signals in the C-band? at the moment controlled by military only while flying over those designated For Galileo that could be an interesting operators (in case of the GPS). Such a regions and for selected periods of time. decision, because by moving its PRS sig- move would further facilitate the devel- Such a time-multiplexing could indeed nals to the C-band, many of the great opment of different concepts for civil prove to be interesting one day. Equally limitations in terms of signal and power and military sectors without having to interesting would be to use special C- that Galileo encountered in the L1 band depend on what the other does. In fact, band-emitting satellites with LEO orbits would be history. this would really open the possibility of to cope with the problem of signal power The idea of moving military signals true interoperability of GNSS system loss and navigation data transmission. and governmental protected services to control segments. Another area where C-band could the C-band would consequently mean Let us reflect for a moment on the play an interesting role is in indoor posi- that the L-band would be then reserved pseudolite concept of pulsing. The rea- tioning and navigation, which is becom-
72 InsideGNSS may/june 2007 www.insidegnss.com working papers
Authors Bernd Eissfeller is a full ing a driving issue for new applications professor of navigation “Working Papers” explore for present and future positioning and and vice-director of the the technical and scien- navigation systems. The attenuation in Institute of Geodesy and tific themes that underpin the C-band is expected to be larger than Navigation at the GNSS programs and appli- in the L-band. Moreover, worse scatter- University FAF Munich. He cations. This regular col- is responsible for ing effects are expected to be observed umn is coordinated by in this band. In fact, in typical indoor teaching and research in Prof. Dr.-Ing. Günter navigation and signal processing. Till the end of environments we can find many objects Hein. 1993 he worked in industry as a project manager that the navigation signals have to go Prof. Hein is a member of the European Com- on the development of GPS/INS navigation through which have a comparable size mission’s Galileo Signal Task Force and organizer systems. He received the Habilitation (venia as the wavelength of the signals. of the annual Munich Satellite Navigation Sum- legendi) in Navigation and Physical Geodesy in On the other hand, the delay spread mit. He has been a full professor and director of 1996 and from 1994-2000 he was head of the that results from these scattering effects the Institute of Geodesy and Navigation at the GNSS Laboratory of the Institute of Geodesy and is longer in the C-band, and at the University of the Federal Armed Forces Munich Navigation. moment a real understanding of the C- (University FAF Munich) since 1983. In 2002, Thomas Pany has a Ph.D. he received the United States Institute of Navi- in geodesy from the Graz band channel for indoor use is still miss- gation Johannes Kepler Award for sustained and University of Technology ing. Could C-band be the indoor band significant contributions to the development of and a M.S. in physics from of the future? satellite navigation. Hein received his Dipl.-Ing the Karl-Franzens Additional Resources and Dr.-Ing. degrees in geodesy from the Univer- University of Graz. sity of Darmstadt, Germany. Contact Prof. Hein at Currently he is working at Hein, G. W. , and J.-A. Avila-Rodriguez, S. Wallner,
www.insidegnss.com may/june 2007 InsideGNSS 73