A Proposed Signal Design for GNSS2: The Use of Faster and Longer Codes to Provide Real-Time Single Frequency Ionospheric Measurements

The design of GNSS2 presents the international navigation community a unique opportunity to design a signal for more precise positioning than is currently available with GPS and GLONASS. Faster and longer acquisition codes would improve code positioning accuracy and signal cross-correlation, as well as providing a greater number of unique codes for pseudolites. Furthermore, receivers may be able to exploit the dispersive nature of the ionosphere and make single frequency measurements, rather than modeling this large source of error. This paper will discuss the numerous advantages of using substantially faster (40-80 Mbps) and longer (16-256 Kbits) acquisition codes for GNSS2. Sample plots will demonstrate that the ionosphere introduces both amplitude and phase modulation for large bandwidth signals, and that these perturbations may permit single frequency ionospheric corrections. GPS and GLONASS have led to a revolution in positioning and navigation. While both systems provide basic accuracy within 100 meters, it should be possible to provide basic accuracy less than 10 meters, after the elimination of ionospheric errors and Selective Availability (GPS only). Currently, stand alone users rely on a simple model of the ionosphere to remove 50-60% of the ionospheric range error. Authorized military users of GPS can make dual frequency measurements and remove more than 90% of the error, depending on the accuracy of the inter frequency bias calibration. Spatial decorrelation and localized scintillation even effect differential users, despite receiving ionospheric corrections. All users would benefit from a system that can measure the local ionospheric conditions and compensate without depending on outside differential corrections. The large frequency difference between L1 and L2 (~350 MHz) permits dual frequency measurements of the ionosphere. This implies that ionospheric measurements should be possible with one high-bandwidth signal. This hypothesis was verified using a computer simulation developed to examine the time domain effects of the ionosphere on a modified GPS signal. A hypothetical GNSS2 (BPSK circularly polarized) signal was synthesized in the frequency domain. The standard model of the ionosphere was used to generate phase corrections equal to -80.6*π*TEC/f. The 'modified' signal was then transformed to the time domain and compared to a 'direct' signal. The results of this simulation show that both phase and amplitude modulation were introduced by the ionospheric model, and that these effects grew according to the square of the bandwidth. Unfortunately these effects are negligible for current GPS and GLONASS signals (code ≤ 10 Mbps). Time domain plots from this simulation show that even with white noise, these modulations should be measurable for faster signals. This paper presents new ways to examine and understand the effect of the ionosphere on a GNSS signal. Single rectangular pulses are examined after passing through the ionosphere. This general case allows a simple method of assessing the distortion. The rectangular pulse is convolved with a short series of impulses to reproduce results for an arbitrary pattern. Finally, noise is added to show that the amplitude and phase modulations can still be recovered. This paper demonstrates that significantly faster (40-80 Mbps) and longer (16-256 Kbits) acquisition codes transmitted at a single frequency will produce five clear advantages: 1) reduced pseudorange variance, 2) improved crosscorrelation, 3) a greater number of PRN codes for pseudolite use, 4) all transmission power in one signal (simplified electronics), and 5) the potential for stand alone users to measure the ionosphere and generate accurate corrections. GPS + GLONASS + EVOLUTION => GNSS2 The design and deployment of the next generation Global Navigation Satellite System (GNSS) will be a challenge in many ways. It will require substantial international cooperation and understanding to develop an improved constellation, signal, and monitoring stations, which in turn will produce improved coverage and accuracy. While this paper will review some previous research, it will concentrate on the signal design for GNSS2. A well-designed signal will be the core of GNSS2, and substantial improvements in accuracy for stand alone users will reduce the dependence on differential corrections. GPS was designed by the U.S. DoD [2] in late 1973, primarily for military use. The signal is transmitted at two frequencies to make ionospheric measurements, and a degraded (civilian) channel to improve signal acquisition. ________ Presented at GNSS-97, 21-24 April 1997, Munich Germany. by J. ChristiePhone (415) 723-9349 Fax (415) 725-9167 email [email protected] Russia has also developed a satellite navigation system [3], known as GLONASS, for their military. While these two systems are quite similar, there are some significant differences that prevent receivers from easily combining measurements from the two constellations. One important difference is the lack of a standard coordinate transformation between the two reference frames. Secondly, the time standards are both based on Universal Time Coordinated (UTC), but there is a bias of milliseconds. One clear advantage of GLONASS is the more highly inclined constellation which improves coverage in high latitude regions, such as Russia. One disadvantage of GLONASS is the use of FDMA which requires more spectrum than CDMA which is used in GPS. The following tables reveal some of the similarities and differences between GLONASS and GPS. Some proposed attributes for GNSS2 are also listed. GPS Civilian Clear Acquistion GPS Military P / Y Code GLONASS GNSS2 Center Frequency (MHz) L1 = 1575.42 L1 = 1575.42 L2 = 1227.60 L1 = 1602+0.5625*N L2 = 1246+0.4375*N N = {1...24} Low end of L band ~ L2 Multiplexing CDMA FDMA CDMA Signal Bandwidth 20 MHz 10 MHz 80 MHz ? Chipping Frequency 1.023 Mbps 10.23 Mbps 511 kbps ~5 Mbps ~40 Mbps Code Length 1023 chips 6.18e+12 chips 511 Repetition Time 0.001 second 1 week 0.001 second >= 0.001 seconds Selective Availability L1 = Yes L2 = No No No Reference Frame WGS-84 (ECEF) PZ-90 (ECEF) ECEF Time Reference US Naval Observatory UTC (no Leap Seconds) Russian UTC UTC Table 1 Comparison of Signals from GPS and GLONASS GPS Civilian Clear Acquistion GPS Military P / Y Code GLONASS GNSS2 # of SV & Period 21 (+3) ~ 11h56min 24 ~ 11h15min 24 ~ 12h 6 ~ 24h Orbital Planes 6 3 6 ? Orbital Inclination 54.0 degrees 64.8 degrees MEO SV > 60 degrees GEO SV ~ 0 degrees Nominal Accuracy ~100 meters w/ SA ~ 25 meters w/o SA ~15 meters ~ 25 meters ~ 5 meters Table 2 Comparison of Constellations from GPS and GLONASS OBVIOUS ADVANTAGES OF LONGER AND FASTER CODES Several of the advantages of using longer and faster codes are based on fundamental ideas of signal detection. Longer codes offer improved cross-correlation properties. By lengthening the linear shift registers from 10 blocks to 18 or 26 blocks, the maximum code is lengthened from 1023 bits to 262,143 chips or 67,108,863 chips. For codes of 1023 bits, the worst case cross-correlation (zero Doppler shift) has 23.8 dB of isolation [2]. Increasing the number of shift registers to 18, produces a total of 48.2 dB of isolation. 26 registers would produce 72.2 dB of isolation, 48 dB more than when using only 10 shift registers. The use of longer codes would also provide more codes for alternate ranging sources. This would help prevent the same PRN code from being transmitted by two different pseudolites both visible from an aircraft, a potentially confusing and dangerous situation. Faster codes are clearly desirable since code tracking is roughly accurate to 1% of a code chip. The precision of GPS C/A code is approximately 3 meters. Whereas for P code, we would expect precision approximately 0.3 meters. However, extremely fast codes (≥ 100 Mbps) require excessive bandwidth will make measurements with precision of approximately 0.03 meters, when the noise is approximately 0.5 meters. However, one disadvantage of longer codes is longer acquisition time. An increase in code length from 1023 bits to 262,143 bits, implies that acquisition times could be lengthened from 30 seconds to 2 hours, which is clearly unacceptable. GNSS2 could solve this problem the same way that GPS does, using a two signals. The basic signal can be quickly and easily acquired, and is used to bootstrap acquisition of the longer more precise codes that cannot be directly acquired. PREVIOUS RESEARCH A previous paper [1] discussed how a substantially faster code chipping frequency could be used to observe and measure the dispersive ionosphere. It showed that the ionosphere introduces both phase and amplitude modulation, but that it is much more pronounced with larger bandwidth signals. However, this method was quite limited, and examined the effect of the ionosphere on only one particular code pattern. Figures 1, 2, and 3 are reproduced from that paper. Figure 1 shows the basic components of the simulation. The user specifies characteristics such as PRN number, code length, and code chipping frequency; and then a GNSS signal is synthesized in the frequency domain. Phase corrections are added based on the standard model of the ionosphere. The signal is then converted to the time domain and analyzed. This simulation focuses on the third and fourth order phase deviations, and removes the first and second order effects that produce 'code-carrier divergence.' This paper will present several innovations that permit a more generalized examination of the effect of the ionosphere on a widely spread signal. Figure 2 shows the amplitude modulation caused by the ionosphere. The upper two panels show the real amplitude of the signal with and without ionosphere, after down converting to baseband. The ionosphere reduces the peak excursion of the signal and slows the bit transitions. The lower panel shows the difference of the two signals. The difference signal has smaller magnitude in longer code blocks (near cycle 100) and is larger near the vertical hash marks which denote changes in code polarity. Note that this plot only shows the real amplitude. The signal after the ionosphere has a significant imaginary amplitude, which appears as phase fluctuations. -3 0 3 Phase Error with (-) and without (:) Ionosphere -200 -150 -100 -50 0 50 100 150 200 -3 0 3 Carrier Cycles (19.0 cm) Phase Error Difference (Code = 112.530 MHz) SIM 14-Sep-96 PLOT 19-Ap -97 Figure 3 Phase modulation due to ionosphere. Figure 3 shows the phase modulation caused by the ionosphere. The phase of the signal is given by φ = ω t, which increases linearly with time, and changes by 180 degrees with each code reversal. The upper panel shows the phase of the signal after subtracting φ(t). The phase modulation of the original signal is nearly zero, with small perturbations near the bit flips, due to filtering effects. The phase of the signal after passing through the ionosphere fluctuates substantially. The lower panel shows the difference of the two signals, and the perturbations are again clustered around changes in the signal polarity. RECTANGULAR PULSE It is difficult to make generalized conclusions about the effect of the ionosphere based on a single code pattern, with only a single level of ionospheric activity. For this reason, the computer simulation was modified to examine the effect of single rectangular pulse that is modulated at the carrier frequency, passes through the ionosphere, and is then demodulated. gnss(t) gnss+iono (t) GNSS(f) GNSS +iono (f) code (t) GNSS (f) gnss (t) Variables code+iono (t) IONO(f) F{} FILT(f) PRN # 1 @ 1.023 Mbps Center Freq = 1.575 MHz Total Electron Count (TEC)