Evaluation of solar radio bursts' effect on GPS receiver signal tracking within International GPS Service network

Authors


Abstract

[1] The direct interference from solar radio bursts (SRB) has not usually been considered as a potential threat to global positioning system (GPS) signal tracking, since the flux densities of most bursts are below 40,000 solar flux units (sfu), a threat threshold to GPS L1 frequency proposed by Klobuchar et al. (1999). Recent analysis indicated that a much lower threshold should be adopted for codeless or semicodeless dual-frequency GPS receivers. In this investigation, severe signal corruptions were found at dayside International GPS Service GPS receiver stations during a large solar radio burst that accompanied the super flare of 28 October 2003. Almost no GPS L2 signals were tracked during the solar flux peak time for areas near the subsolar point. Correlation analysis was performed between the rate of loss of lock on GPS L2 frequency and solar radio flux density at different bands, and a correlation index as high as 0.75 is revealed in the 1415 MHz solar radiation band, which is located between the two GPS operating frequencies L2 (1227.60 MHz) and L1 (1575.42 MHz). The correlation analysis indicates that GPS signal losses of lock were primarily caused by microwave in-band interference and that the threat threshold of SRB effects on the GPS system should be re-evaluated, since the flux density of the burst at 1415 MHz was just 4,000–12,000 sfu, which is far below the previously proposed threat threshold. The signal-tracking performance of different types of GPS receivers during such a super flare event is also presented.

1. Introduction

1.1. Solar Radio Burst Interference to GPS

[2] There are several different kinds of interference sources to global positioning system (GPS) signals, such as in-band emission, nearby-band emission, harmonics, and jamming, which may potentially disrupt a GPS receiver's signal tracking [Johannessen, 1990; Owen, 1993; Moelker, 1994; Stevens, 1995]. In most cases, such interference comes from radio communications, mobile phones, power lines, radar systems and equipment operated by police and emergency vehicles etc. [Johannessen, 1990; Owen, 1993; Moelker, 1994; Ward, 1994]. Radio frequency interference (RFI) can decrease the GPS signal-to-noise ratio (SNR) by the introduction of additional noise that subsequently leads to degradation in GPS positioning accuracy or a loss of GPS signal in the worst case.

[3] Solar radio bursts (SRB), a source of radio frequency interference, have been studied for more than 50 years [Reber, 1944; Southworth, 1945]. SRB occur in the solar atmosphere with three different types of radio emission mechanisms, whose frequency band ranges over the entire radio spectrum [Reber, 1944; Kundu, 1965; Castelli et al., 1973; Guidice and Castelli, 1975]. The interference of SRB at radio frequencies was first reported by Hey [1946], who noticed that interference occurred during solar flares. Recent studies also showed that SRB in the spectrum of microwave radio frequencies can disrupt wireless communications, where the threshold was set to about 1000 solar flux units (sfu) (1 sfu = 10−22 W m−2 Hz−1) [Bala et al., 2002]. SRB with flux density above this threshold typically occur once every 3.5 days during maximum solar activities [Bala et al., 2002].

[4] GPS antennas are designed to be right-hand circularly polarized (RHCP) according to GPS signal RHCP characteristics. Interference signals that do not match the GPS antenna polarization pattern will be reduced in strength dependent upon the degree of mismatch. Klobuchar et al. [1999] has investigated the threshold of radio frequency interference on L1 frequency (1575.42 MHz) due to SRB for GPS signals, with the receiver background noise level (or thermal noise density) set as −201.5 dBW/Hz and the antenna gain as 1 dB. The obtained threshold of SRB interference to phase tracking loop operating at L1 frequency was 40,000 sfu for a randomly polarized solar radio emission and 20,000 sfu for a RHCP emission [Klobuchar et al., 1999]. In the past 40 years, only a few solar radio bursts were observed with peak flux density over 40,000 sfu [Klobuchar et al., 1999; Bala et al., 2002; Nita et al., 2002]. Therefore solar radio bursts were thought to rarely become a potential interference source to GPS signals. In this investigation, we have found that SRB with a flux density of only 12,000 sfu can cause a severe interference to GPS signal tracking.

[5] In recent years, dual-frequency GPS receivers have been widely employed to estimate ionospheric total electron content (TEC). In order to extract the encrypted GPS L2 signal, codeless or semicodeless technologies are widely used in GPS receivers, which however make GPS L2 signal much more prone to interference [Skone, 2001]. To assess the threshold of GPS L2 signal tracking induced by the correlation-tracking mode employed by dual-frequency receivers, the following equation (1) is derived in our research (according to Spilker [1996], Ward [1999], and Klobuchar et al. [1999]):

equation image

where Pthr is the threshold of GPS L2 signal tracking in dB (corresponding to y-axis value in Figure 1); Sr is the power of GPS L2 signal (this value is doubled here since GPS signal power was compared with both interference signal and receiver thermal noise); Nthermal is the receiver background thermal noise density; Lbw is the bandwidth loss on SNR due to bandwidth of GPS signal power spectrum and RF circuit design; A is the antenna effective area in dBmr2; SNRcorr is the correlation gain which is decided by hardware design in order to lock GPS signals; dBLoss is the SNR loss in L2 signal tracking determined by the receiver-tracking mode (codeless or semicodeless, this value corresponds to x-axis value in Figure 1 with a unit of dB); SNRPLL is the SNR detection threshold for phase lock loop used in the GPS receiver; −220 dBW/Hz is the unit transform constant since 1 sfu = −220 dBW/Hz. Shown in Figure 1 are the calculated thresholds (in unit of sfu) of SRB effects on GPS L2 signal using equation (1). To generate the plot, the following values have been applied: Nthermal = −203 dBW/Hz which is a typical value widely used for the analysis of RFI effects [Ward, 1999]; A = 10 log(gλ2/4π) with g = − 3 dB for randomly polarized radiations; Sr = −163 dBW; Lbw = 10 × log10 (500 × 103 Hz) since most L2 signal power is concentrated on the narrow bandwidth region of 500 kHz; SNRcorr = 10 × log10 (1.023 × 106) = 60 dB for 1 ms integration and SNRcorr = 10 × log10 (4 × 1.023 × 106) = 66 dB for 4 ms integration, both computed with respect to a P code chip length of 1.023 × 106 bits; SNRPLL is typically set as 14 dB; As denoted with vertical lines in Figure 1, dBLoss is between 14∼17 dB for semicodeless receiver and between 27∼30 dB for codeless receivers.

Figure 1.

Thresholds of SRB effects on GPS L2 signal.

[6] According to the results shown in Figure 1, the threshold is about 1.0 × 106∼4.0 × 106 sfu for full code correlation of GPS L2 signal, about 2.0 × 104∼1.0 × 105 sfu for semicodeless receivers and about 1,000∼8,000 sfu for codeless receivers. The threshold therefore varies for different types of receivers.

1.2. SRB on 28 October 2003

[7] A strong solar radio burst was observed recently on 28 October 2003 and the maximum X-ray flux X17.2 occurred at 1104 UT as recorded by Geosynchronous Operational Environment Satellites (GOES). The data from the National Geophysical Data Center (NGDC)'s Radio Solar Telescope Network (RSTN) and some other European solar astronomy observatories showed that the SRB on microwave band mainly occurred at two time periods on 28 October 2003, namely 1102∼1112 UT and 1142∼1200 UT [National Geophysical Data Center, 2004; Swiss Federal Institute of Technology, PHOENIX-2 ETH, available at http://www.astro.phys.ethz.ch/rapp/catalog/catalog_nf.html#phoenixII; Astronomical Institute, Academy of Sciences of the Czech Republic, Solar radio event archive info, available at http://sunkl.asu.cas.cz/∼radio/info.htm]. The frequency band of SRB that we are concerned about in this paper is the microwave one where GPS frequencies are located. In the first period, the peak value of the solar flux at 1415 MHz reached 7,000 sfu. It reached 12,000 sfu in the second period. During both periods, losses of tracking to GPS signals have been observed by many International GPS Service (IGS) dayside stations. A correlation analysis on the rate of loss of lock at solar radio flux at 1415 MHz indicated that, taking GPS station NKLG as an example, it had a correlation index r = 0.75 which is much higher than correlation indices in other frequency bands. In section 2, we will provide a global distribution of the tolerance conditions for the IGS network to this event during the first time period (1102∼1112 UT). In section 3, a global map of loss of lock will be generated for periods 1051∼1100 UT and 1112∼1116 UT (before and after the peak solar flux), corresponding to the pre-effects and posteffects of the SRB on GPS receivers, respectively.

[8] During the super flare, with sudden enhanced X-ray flux from the Sun, the ionospheric TEC experienced a sudden enhancement due to extra ionization by solar X-ray and extreme ultraviolet (EUV) flux. The largest TEC enhancement occurred at equator dayside regions with a value reaching 15 TEC units, (TECU) (1 TECU = 1016 el m−2). Large sudden TEC enhancements will cause loss of lock on GPS L2 signal tracking as well [International GPS Service (IGS), 1998, 1999a, 1999b]. A sudden TEC enhancement was observed during 1102∼1106 UT, and it has four minutes overlap with the time period 1102∼1112 UT during which GPS signal loss of lock was noticed. Therefore TEC sudden enhancement could be another source affecting L2 signal tracking since different types of GPS receivers have used different tracking techniques to acquire GPS L2 signal and their capability in SRB interference resistance will be different. Considering the above, different types of GPS receivers should be analyzed in the evaluation of SRB's effects on GPS L2 signal tracking.

2. Observation and Data Analysis

2.1. Disruption of GPS Phase Tracking During SRB

[9] Shown in Table 1 is the GPS L2 signal-tracking status for all the visible satellites at IGS station NKLG (Geographic Lat/Lon: 2.1°/9.4°, TRIMBLE 4000SSI receiver) near the subsolar point during 1102∼1159 UT on 28 October 2003. For each satellite, which is identified by its pseudo-random number (PRN), the status is marked with “OK” if the L2 signal was tracked otherwise it is marked with “L2” if the L2 signal tracking was lost. It can be seen from Table 1 that severe loss of lock on GPS L2 signal occurred at NKLG during that period. At the epoch 1105:00–1105:30 UT, the tracking to all visible satellites was totally lost and the rate of loss of lock was 100%.

Table 1. Loss of Lock on L2 Phase Tracking During 1102∼1159 UT for GPS Station NKLG
TimePRN5PRN10PRN7PRN17PRN30PRN9PRN24PRN18PRN29PRN26
1102OKOKOKOKOKOKOKOKOK
1103OKOKL2L2L2L2L2OKL2
1104OKOKL2L2L2L2L2OKL2
1105L2L2L2L2L2L2L2L2L2
1106OKOKL2L2L2L2L2OKOK
1107OKOKOKL2L2L2L2OKOK
1108OKOKOKL2L2L2L2OKOK
1109OKOKOKOKL2OKL2OKL2
1110OKOKOKOKOKOKL2OKOK
1140OKOKOKOKOKOKOKOKOK
1141OKOKOKOKOKOKOKOKOK
1142OKOKOKOKOKOKOKOKOK
1143OKOKL2OKOKL2L2OKOK
1144OKOKL2L2OKOKL2OKOK
1145OKOKL2L2L2OKL2OKOK
1146OKOKL2L2L2L2L2OKOK
1147OKOKL2L2L2L2L2OKOK
1148OKOKL2L2L2L2L2OKOK
1149OKOKL2OKOKOKL2OKOK
1150OKOKL2OKOKOKOKOKOK
1151OKOKL2OKOKOKOKL2OKOK
1152OKOKL2OKOKOKL2OKOK
1153OKOKL2OKOKOKL2L2OKOK
1154OKOKL2L2OKOKL2L2OKOK
1155OKOKL2L2OKOKL2L2OKOK
1156OKOKL2OKOKL2OKL2OKOK
1157OKOKL2L2L2L2L2L2OKOK
1158OKOKL2L2L2L2L2L2OKOK
1159OKL2L2L2OKL2OKL2OKOK

[10] Similar loss of lock on GPS L2 signal tracking was also observed at other dayside GPS stations at a global scale such as BRAZ (geographic latitude −47.9°, longitude −15.9°), ALAC (geographic latitude +38.2°, longitude −0.3°) and SIMO (geographic latitude 18.4°, longitude −34.2°). The statistics of the GPS L2 signal loss of lock are illustrated in Figure 2.

Figure 2.

Rate of loss of lock on L2 phase tracking during 1050∼1200 UT, 28 October 2003, at several IGS stations.

[11] Data for solar radio flux collected by NGDC were recorded at 1-s rate and at frequencies 245 MHz, 410 MHz, 610 MHz, 1415 MHz, 2695 MHz, 4995 MHz, 8800 MHz and 15400 MHz. The solar radio flux data were resampled at 30-s rate in order to match the 30-s GPS data rate used by GPS receivers within the IGS network. Shown in Figure 3 are the correlation indices between the solar radio flux at different frequencies and the rate of loss of lock at each epoch during 1048:00∼1230:00 UT at the station NKLG. In Figure 3, the solid line in each plot gives the least squares linear fit with the correlation index shown at the top of each plot, the y axis is the flux density of solar radio with a unit of sfu and the x axis is the rate of loss of lock with a unit of %. As seen from Figure 3, the correlation index between the solar radio flux at the frequency 1415 MHz and the rate of loss of lock reached 0.75, which was much higher than the correlation indices at other frequencies. A correlation index of 0.60∼0.80 is usually considered to indicate a remarkable degree of correlation [Franzblau, 1958]. Considering the fact that the frequency 1415 MHz is located between GPS L1 frequency (1575.42 MHz) and L2 frequency (1227.60 MHz) and that the frequencies of solar radio bursts are continuous in the microwave band, the maximum correlation at the frequency of 1415 MHz implies that the in-band interference caused by solar radio flux at the GPS signal frequencies has a more prominent capacity than other frequencies to corrupt GPS L2 signal tracking.

Figure 3.

Correlation between the rate of loss of lock and the solar radio flux measured at different frequency bands.

2.2. SRB Impacts on the IGS Network

[12] In order to get a global map of SRB effects on the IGS network, data from over 900 GPS stations were downloaded from the IGS analysis center (ftp://garner.ucsd.edu/) and analyzed. We studied the data recorded during the period 1102:00∼1112:00 UT when the first major solar radio burst occurred on 28 October 2003. Within this time period (10 min), if a receiver at a station lost its L2 phase tracking to a satellite, a loss of lock to that satellite is considered to have occurred. A statistic of the losses of lock for the 900 GPS stations is shown in Figure 4. The x axis and y axis represent the latitude and longitude of the stations, respectively, while the z axis shows the PRN of the GPS satellites. Each point in Figure 4 represents a satellite that had been tracked by a GPS receiver. A red point indicates a loss of lock to a satellite (identified by its PRN) occurred at a GPS receiver while a blue point indicates a receiver has maintained its tracking to a satellite. A GPS receiver typically has 12 or more channels so it is able to simultaneously track signals from all visible satellites.

Figure 4.

Loss of lock on L2 frequency during 1102:00∼1112:00 UT, 28 October 2003, using GPS data from the IGS global network.

[13] Shown in Figure 4 are the losses of lock on L2 signal at global GPS stations. From Figure 4, however it is difficult to identify the characteristics of the geographical distribution of the loss of lock. In the following we adopt the grid concept to project the global occurrence of the loss of lock to compute the rate of loss of lock (in the unit of percent) in different geographic grids by equation (2):

equation image

where Pi ×j is the rate of loss of lock in the geographic grid io, jo region; i represents the longitude with a range from −180° to 180°; j denotes the latitude with a range from −90° to 90°; (NPLL)ij is the total number of satellites which have experienced loss of lock on GPS L2 signal for stations located in the grid area of io, jo; (NALL)ij is the total number of visible satellites observed by the stations whose positions are within the same grid area.

[14] Shown in Figure 5 is a global map of loss of lock on GPS L2 signal for the time period of 1102:00∼1112:00 UT on 28 October 2003. The local noon time line and the day-night terminator line are shown by the yellow line and curves, respectively. The location of the subsolar point is shown by a solid yellow point. Figure 5 shows that almost all dayside GPS stations were influenced by the SRB. On the night side regions, the loss of lock mainly occurred in near the geomagnetic equator regions and the polar regions.

Figure 5.

Rate of loss of lock for IGS network during 1102:00∼1112:00 UT, 28 October 2003.

3. Discussion

[15] SRB-caused loss of lock on GPS L2 signal mainly occurred in the dayside hemisphere and especially near the subsolar point region, as indicated in Figure 5. GPS data from station NKLG for day of year 301 in 2000, 2001, 2002 and days of year 300 and 302 in 2003 were also investigated for the time period 1100∼1300 UT (local noon with solar overhead) but no interference effects were found. Thus it is confirmed that the loss of lock at a global scale as shown in Figure 5 was caused by the severe solar radio bursts on 28 October 2003.

3.1. Pre-Effects and Posteffects of SRB on the IGS Network

[16] As shown in Figure 5, severe losses of lock on GPS L2 signal are observed during solar radio bursts 1102:00∼1112:00 UT on 28 October 2003. To investigate the preburst and postburst SRB effects on the global GPS network, we study two time periods: namely t1 (1051:00∼1100:00 UT) and t2 (1112:00∼1116:30 UT), which are two time periods before and after the strongest part of the solar radio burst as shown in Figure 6. The preburst and postburst SRB effects on IGS network are depicted in Figure 7 and Figure 8, respectively, using the same statistical methods as used for Figure 5. Figure 7 does not show severe loss of lock on GPS L2 signal at a global scale during the preburst period 1051:00∼1100:00 UT. Only about 10% tracking channels have been corrupted around the subsolar point regions. Figure 8 shows that postburst SRB has a reduced effect on GPS stations. Although losses of lock on GPS signal have been observed at some GPS stations in some areas, it is not a global phenomenon. The rate of loss of lock is 50% for the near subsolar point region, which is significantly lower than 90% during solar radio bursts period 1102:00∼1112:00 UT in the same area.

Figure 6.

Sketch plot (not linear) of timescale used in this study. Time in universal time is denoted below the arrow line, and the timescale of ionospheric response to solar X-ray and EUV flux (sudden TEC enhancement) is given; solar X-ray class is given above the arrow line, denoted with timescale of loss of lock and t1 and t2.

Figure 7.

Rate of loss of lock for IGS network before burst (L2 signal, 1051∼1100 UT).

Figure 8.

Rate of loss of lock for IGS network after burst (L2 signal, 1112:00∼1116:30 UT).

3.2. Sudden Ionospheric TEC Enhancement

[17] Solar short-wave bursts on X-ray band and extreme ultraviolet band caused extra ionization in the plasmasphere (http://image.gsfc.nasa.gov/), and the ionospheric D, E and F regions [Garriott et al., 1967; Zhang and Xiao, 2003]. Compared to historical records, the flare on 1100 UT, 28 October 2003, is the third largest one (X17.2) in the history. During 1102:00∼1106:30 UT on 28 October 2003, the solar flare was observed to cause a sudden ionospheric TEC enhancement as shown in Figure 9. During this flare, the ionospheric TEC enhancement lasts 10 min at a global scale synchronously on the dayside, and in some area the increased value even reached 15 TECU. This may contribute to the loss of lock on GPS L2 signal as well [IGS, 1998, 1999a, 1999b]. This type of loss of lock primarily occurs to GPS receivers that employ cross-correlation tracking mode (e.g., AOA TurboRogue receivers within the IGS network), and this effect is centered in subsolar point region. From the analysis of GPS L2 signal loss of lock, it is difficult to discriminate the effect of sudden TEC enhancement and that of SRB in-band interference because they occur during the same period. However, it can be concluded that the sudden TEC enhancement is a secondary factor causing the GPS signal loss of lock. This is because the sudden TEC enhancement mainly affected the GPS receivers that employ cross-correlation technique to acquire the L2 signal. Within the IGS network, only a small portion of GPS receivers such as TurboRogue receivers employ such a technique. That is to say the sudden TEC enhancement would cause loss of lock only to the GPS stations equipped with cross-correlation receivers rather than to all GPS stations. Therefore the solar radio bursts are the primary source that has degraded the IGS network tracking performance during the super flare. The ionospheric TEC enhancement plays only a secondary role for the presence of the losses of L2 signal tracking.

Figure 9.

Sudden TEC enhancement caused by extra ionization in ionosphere and plasmasphere by solar X-ray and EUV bursts during 1102∼1106:30 UT, 28 October 2003. Color bar gives the numerical value of enhanced TEC in units of TECU (1 TECU = 1016 el m−2).

3.3. Relationship Between Receiver Type and L2 Phase Loss of Lock

[18] The type of GPS receivers and antennas is another key factor that affects the capability to resist SRB in-band interferences on GPS signal tracking and acquisition. GPS manufacturers currently employ different tolerance thresholds to RFI in the process of antenna design, interference rejection, code and carrier signal processing, and demodulation of navigation message. Figure 10 illustrates a comparison of the tracking performance of four different types of GPS receivers at the dayside GPS stations. ASHTECH Z-XII3 is denoted by the diamonds, TRIMBLE 4000SSI by the circles, AOA SNR-8000 ACT by the squares, and AOA ROGUE SNR-8000 by the stars. The red color in Figure 10 indicates that L2 signal was corrupted during 1102∼1112 UT, and the blue color indicates no corruption. Figure 10 shows that the GPS L2 signal tracking at stations using TRIMBLE 4000SSI and ROGUE SNR-8000 receivers are almost all corrupted. Table 2 summarizes the occurrence of the GPS L2 signal phase loss of lock during SRB for major GPS receiver types in the dayside regions. A total of 156 stations were used in the analysis. Table 2 indicates that TRIMBLE 4000SSI and AOA ROGUE SNR-8000 were more vulnerable during this event to solar radio burst interference than ASHTECH Z-XII3, UZ-12 and Z18 receivers and AOA BENCHMARK ACT and SNR-8000 ACT receivers.

Figure 10.

Comparison of dayside GPS receiver operation within four types of GPS receivers during 1102∼1112 UT, 28 October 2003.

Table 2. Relationship Between Receiver Types and Phase Lock Loop (PLL) Error Occurrence
Receiver TypesTracking ModeStation Number (Total 156)Percent of Stations With PLL (1102∼1112 UT)Percent of Stations With PLL (1142∼1200 UT)
ASHTECH Z-XII3codeless4311.622.33
ASHTECH UZ-12codeless128.330.00
ASHTECH Z18codeless1127.2718.18
TRIMBLE 4000SSIcross correlation3878.9563.16
TRIMBLE 4700codeless922.2222.22
TRIMBLE 4000SSEcross correlation616.6716.67
AOA SNR-8000 ACTcodeless147.147.14
AOA BENCHMARK ACTcodeless80.000.00
AOA ROGUE SNR-8000cross correlation875.0075.00
JPS LEGACYcodeless714.2914.29

4. Conclusions

[19] Severe losses of lock on GPS L2 signal were observed at GPS stations in the global IGS network during the solar radio burst event on 1102∼1112 UT and 1142∼1200 UT, 28 October 2003. A peak correlation r = 0.75 was revealed between GPS L2 tracking corruption and 1415 MHz solar radio flux. The study showed that solar radio bursts with a density of 4,000∼12,000 sfu, which is much lower than current threat threshold density 40,000 sfu widely used for previous investigations, could also cause a severe impact on GPS L2 signal tracking. The results indicate that the determination of a more appropriate threshold for solar radio bursts effect on GPS signals should be examined.

[20] The pre-SRB and post-SRB effects have also been investigated in this study and it was revealed that solar radio bursts are the dominant interference source that caused severe losses of lock of GPS L2 signal as to the 28 October 2003 event. The ionospheric TEC sudden enhancement played only a secondary role in causing L2 signal loss of lock. The tracking performance of different types of GPS receivers under this solar radio burst event was also presented.

Acknowledgments

[21] This research was supported by a strategic grant from Natural Sciences and Engineering Research Council of Canada (NSERC).

Ancillary