A Statistical Analysis of GPS L1, L2, and L5 Tracking ... · PDF fileA Statistical Analysis of GPS L1, L2, and L5 ... Institute for Scientific Research, Boston College, ... Brazil,

Proc. 2015 ION Pacific PNT Conference 1 Honolulu, Hawaii, April 20-23, 2015

BIOGRAPHY

Susan H. Delay is a senior research analyst at Boston

College’s Institute for Scientific Research. She

received a B.A. in Mathematics from Trinity College

and an MA from Boston College. Her interests are

space weather, satellite communications and the

Global Positioning System.

Charles S. Carrano is a senior research physicist at

Boston College’s Institute for Scientific Research. He

received a B.S. in mechanical engineering from

Cornell University and M.S. and Ph.D. degrees in

aerospace engineering from The Pennsylvania State

University. His research interests include ionospheric

impacts on radar, satellite communications, and the

Global Positioning System.

Keith M. Groves is currently a Program Manager in

the Space Vehicles Directorate of the Air Force

Research Laboratory where he investigates

ionospheric scintillation and its impact on satellite

based communication and navigation systems. He has

a Ph.D. in Space Physics from MIT and a B.S. in

Physics from Andrews University.

Patricia H. Doherty is the Director and a Senior

Scientist of the Institute for Scientific Research (ISR)

at Boston College (BC). As director of the Institute,

she oversees the activities of staff members working

on a variety of innovative research projects. As a

scientist, Patricia’s own research interests are

centered on the ionospheric effects on Satellite-Based

Augmentation Systems (SBAS) and on promoting

research and education in the science of navigation in

developing countries.

ABSTRACT

GNSS navigation accuracy in the presence of

ionospheric scintillation depends critically on

tracking loop performance, which can be

characterized in terms of the probability of loss-of-

lock (LOL) and the time for signal reacquisition

following LOL events. Due to the relatively recent

introduction of the new GPS modernization signals

L2C and L5, there have been few statistical studies

comparing L1, L2C, and L5 tracking performance

under real-world scintillation conditions. While the

lower frequency carriers generally experience larger

signal fluctuations (due to the well-known frequency

dependence of scintillation), the different codes and

tracking algorithms employed for the different carrier

signals make it difficult to predict their vulnerabilities

and potential benefits to the NextGen aviation

systems that will leverage these signals. Moreover,

different GNSS receiver models employ different

tracking algorithms which may exhibit unique

strengths and vulnerabilities, depending on the

characteristics of the scintillating environment. The

most direct way to assess the tracking performance

for L1, L2C, and L5 during scintillation is by the

statistical analysis of experimental data collected

using multiple receiver models during the current

solar maximum period.

With funding and support from the Federal Aviation

Administration (FAA), Boston College and National

Institute for Space Research (INPE) have been

collecting GNSS scintillation observations in Brazil

since 2012. Both GPS legacy and triple frequency

receivers (L1 C/A, L2C and L5) are represented. As

part of our ongoing study to assess GNSS signal

tracking performance and navigation accuracy during

scintillation, in this paper we report on the probability

of losing code lock on L1, L2C, and L5 with two

widely used GNSS scintillation monitors, the

NovAtel GPStation-6 and Septentrio PolaRxS Pro.

The approach we have taken is to count the number

of scintillation-induced gaps in the high rate (50 Hz)

receiver-reported signal amplitudes. Next, we bin

these data gaps as a function of the scintillation index

S4. The ratio of the number of missing samples to the

total number of samples for a given S4 yields the

probability of interrupted code tracking as a function

of S4. Only high elevation observations are included

in the statistics to exclude fluctuations caused by

multipath reflections from terrestrial objects.

A Statistical Analysis of GPS L1, L2, and L5

Tracking Performance During Ionospheric

Scintillation Susan H. Delay, Charles S. Carrano, Keith M. Groves, Patricia H. Doherty

Institute for Scientific Research, Boston College, Chestnut Hill, MA


We find that both receivers the NovAtel GPStation-6

and Septentrio PolaRxS Pro generally experienced a

higher probability of losing lock on the lower

frequency carriers (L2C and L5), even when

quantified in terms of S4 on the same carrier, despite

the enhanced codes and advanced tracking techniques

available for these modernization signals.

INTRODUCTION

Ionospheric scintillations are fluctuations in the

intensity and phase of satellite signals caused by

scattering from electron density irregularities in the

ionosphere. The intensity of scintillations is

positively correlated with the solar cycle and the

associated signal fades often exceed 20 dB at L-band

frequencies during solar maximum. Scintillation is

generally most intense in the equatorial region of the

globe after sunset, but it also occurs in the northern

and southern high latitude regions. The occurrence

morphology of scintillation depends on season,

longitude, solar cycle, magnetic activity, and exhibits

a high degree of night-to-night variability (Aarons

1982; Aarons 1993).

Ionospheric scintillation affects GPS receivers in

multiple ways. Amplitude scintillations result in

errors decoding the GPS data messages and

estimating the ranges to the satellites. Phase

fluctuations stress the ability of the receiver to

maintain lock on the signals and can cause “cycle

slips” or breaks in the measured phase. These cycle

slips may prevent the receiver from using the phase

to refine its range measurements. When the receiver

is unable to maintain lock on at least four or more

GPS satellite signals, a temporary loss of positioning

service occurs. The duration of these “outages” in

service depends on the duration and severity of the

disturbances, the geometry of the satellites in view,

and the signal reacquisition time of the equipment

(Carrano et al., 2005, Carrano et al., 2010).

The S4 index is the standard deviation of normalized

signal intensity fluctuations and is directly related to

the probability that signal fades will reach a

particular level (Basu et al., 1987). Loss of lock is

more likely to occur when the GPS signal level drops

below the fade margin of the receiver’s internal

tracking loops. Therefore, it is not surprising that the

probability of losing lock varies as a function of the

S4 index (Carrano et al, 2010).

In this paper we explore the statistical relationship

between intensity fluctuations due to ionospheric

scintillation and loss of lock occurrence on the GPS

L1, L2C, and L5 for two widely used GNSS

scintillation monitors. We expect this information

will be useful in future modeling and simulation

studies concerning the impacts of scintillation on

GNSS navigation accuracy.

THE DATA COLLECTION SITE

The data considered in this study was collected at São

José dos Campos, Brazil, a location near the crest of

the Appleton anomaly where global scintillation

levels tend to be strongest (coordinates: S16°54.42' /

W47°42.50'16.9 degrees). Figure 1 shows the

location of the site. We initially selected scintillation

observations between November 2012 and January

2013 for this analysis but, for reasons we shall

discuss, we also considered data during the period

September to October 2014.

Figure 1. The asterisk indicates the location of São

José dos Campos, Brazil. The dashed line shows the

location of the magnetic dip equator.

GNSS RECEIVER HARDWARE

Two different GNSS receiver models were used to

collect the scintillation observations in this project,

namely the NovAtel GPStation-6 and the Septentrio

PolaRxS Pro. Both receivers were co-located at the

site but operated with separate L1/L2/L5 antennas

made by their respective manufacturers. Below, we

provide a brief description of the equipment.

Additional information is available from the receiver

manufacturers’ websites.

NovAtel GPStation-6

This GNSS scintillation monitor can track the GPS

L1/L2/L2C/L5, SBAS L1/L5, GLONASS L1/L2,

Galileo E1/E5a/E5b/Alt-BOC and BeiDou signals.

Signal power and phase measurements are provided

at a sampling rate of 50 Hz. A total of 120

independent channels are available for tracking

signals. Signal intensity measurements are provided

as the difference between narrow band and wide band


power measured over 20 millisecond periods (Van

Dierendonck, et al., 1993; Falletti et al., 2010). An

Oven Controlled Crystal Oscillator (OCXO) is

employed for low phase noise. This receiver has no

direct access to the GPS P(Y) code. It monitors the

open codes on L2C and L5 to produce observations

on those carrier frequencies suitable for scintillation

analysis (Shanmugan et al., 2012).

Septentrio PolaRxS Pro

This Global Navigation Satellite System (GNSS)

receiver can monitor satellites from the following

constellations: GPS, GLONASS, Galileo, and SBAS.

Using 136 channels, it can track the L1, L2, L2C, L5,

and E5ab / AltBOC satellite signals. While this

receiver is capable of providing power and phase

measurements at 100 Hz, we operated it at 50 Hz. We

computed signal intensity from the post-correlator In-

phase (I) and Quadrature (Q) samples acquired

during 20 millisecond intervals. This receiver uses a

high-quality OCXO for low phase noise. This

receiver has no direct access to the GPS P(Y) code. It

monitors the open codes on L2C and L5 to produce

observations on those carrier frequencies suitable for

scintillation analysis (Spoglia et al., 2013).

METHODOLOGY

The approach we have taken is to count the number

of scintillation-induced gaps in the high rate (50 Hz)

receiver-reported signal amplitudes, and then

compute the probability of scintillation-induced loss

of lock as a function of S4. The steps we take to

accomplish this are as follows.

First we attempt to exclude data gaps which may be

caused by processes other than ionospheric

scintillation. These include multipath and receiver

noise. To avoid the former, we discard data from low

elevation satellites (<30) and also data gaps that last

longer than 5 minutes (setting satellites). To avoid the

latter, we consider only data for which S40.3. The

remaining data gaps are considered to be the result of

scintillation-induced loss of lock. We count the total

number of missing samples in these gaps.

Next, we interpolate the S4 index (computed every

60 sec) onto the high rate (50 Hz) data epochs, and

bin the scintillation-induced data gaps according to

S4. The ratio of the number of missing samples to the

total number of samples for each S4 yields the

probability of a scintillation-induced data gap

(interrupted code tracking) as a function of S4.

Both the NovAtel GPStation-6 and the Septentrio

PolaRxS Pro have been specifically designed to

provide robust signal tracking during ionospheric

scintillation. Nevertheless, loss of lock does occur

when the scintillations are sufficiently intense,

leading to gaps in the measured data. Figures 2 and 3

compare 50 Hz L1 signal power fluctuations

measured by the Novatel and Septentrio receivers,

respectively, between 0-4 UT on 16 November, 2012.

Only data for PRNs 21, 22, 24, 25, 26, 29, 30, and 31

are shown (the data for other tracked satellites are

omitted for clarity). Missing data samples (data gaps)

associated with loss of lock events are colored red,

whereas green indicates uninterrupted signal

tracking. Despite the label “C/No” used for the

vertical axes in Figures 3 and 4, the data plotted are

actually signal intensities, plus arbitrary offsets

(determined by us). Hence one should not compare

the absolute magnitudes of the C/No values shown

between the two receivers. Our LOL statistics will be

reported in terms of S4, which is computed from the

normalized signal intensity, so that the absolute

magnitude of signal intensity is unimportant.

A number of specific events have been identified in

Figure 2 for discussion. The events circled in yellow

were recorded while the transmitting satellites were

high in the sky (>30 elevation). Hence the data gaps

(red) that occurred during these events were likely

caused by the signal fluctuations associated with

ionospheric scintillation. Other events circled in red

were recorded while the transmitting satellites were

at low elevation angles. Signal fluctuations associated

with these events are due to multipath, ionospheric

scintillation, or some combination of both. Data gaps

associated with these low elevation events have been

excluded from our statistics.

When we first examined the statistics of interrupted

signal tracking for the PolaRxS Pro receiver, we were

surprised to find many gaps in the data from high

elevation satellites, even in the absence of any

scintillation. This problem is evident in Figure 3,

which shows that there are multiple instances of

simultaneously interrupted tracking events for all

satellites tracked. Clearly this is not an environmental

effect but instead due to some problem with the

equipment. Since the PolaRxS Pro and GPStation-6

receivers were co-located, the problem is unlikely

due to the source of local RF interference. We

conjecture that either the receiver was unable to track

the requested satellite signals due to an overburdened

CPU, or it was unable to send the data to the

computer quickly enough over the manufacturer-

supplied USB to high-speed serial adaptor cable. In

any case, it became clear that we could not use the

data from this particular PolaRx Pro receiver for our

analysis.


Figure 2. L1 signal power for the NovAtel GPStation-6 (not all tracked satellites are shown). Missing data samples

(data gaps) due to loss of lock are shown in red, whereas green indicates uninterrupted signal tracking.

Figure 3. L1 signal power for the Septentrio PolaRxS Pro (not all tracked satellites are shown). Missing data

samples (data gaps) due to loss of lock are shown in red, whereas green indicates uninterrupted signal tracking.


We should note that when the Septentrio receiver was

tracking normally, it tended to experience fewer

scintillation-induced data gaps than the NovAtel

receiver when tracking L1 signals. For example,

compare the scintillation event circled in yellow

toward the upper-right hand corner of Figure 2 with

the corresponding scintillation event in Figure 3.

Thankfully, there is another Septentrio PolaRxS Pro

operating at São José dos Campos (about 10 km

distant from our site), as part of the CIGALA

network (Bougard et al., 2011). João Francisco

Galera Monico kindly provided data from the

CIGALA PolaRxS Pro receiver for the period

September-October 2014. We are very grateful to Dr.

Galera Monico and his team for supplying this data

for use in our study. From this point in the paper

onward, the Septentrio data we will discuss is from

the CIGALA receiver.

When we examined the data from the CIGALA

PolaRxS-Pro receiver, we noted elevated S4 values

throughout the day (as large as 0.5), even at high

elevation angles. A closer examination of the raw

signal intensities revealed evidence of simultaneous

fading (but not loss of lock) for all satellites tracked.

We suspect a source of local RF interference may be

responsible for the elevated S4 values, but thankfully

our LOL statistics should be unaffected, since (as we

shall see) the PolaRxS Pro maintained satellite

tracking on L1 when S4 was less than 0.6.

Figure 4. A comparison of L1 tracking performance between the GPStation-6 (left) and PolaRxS Pro (right).


STATISTICS OF SCINTILLATION-INDUCED

TRACKING INTERRUPTIONS

Next we present the results of our statistical analysis.

Figure 4 compares the results for the GPStation-6

(left) and PolaRxS Pro (right) while tracking the GPS

L1 signal. The format of this figure, and the ones to

follow, is as follows. The top panel is a histogram

showing the total number of 50 Hz data samples

collected as a function of S4. Data for satellites

below 30 elevation and data gaps lasting longer than

5 minutes are excluded from this number. The middle

panel shows the number of missing 50 Hz data

samples as a function of S4. The lower panel shows

the ratio of the latter to the former, which gives the

probability of a scintillation-induced data gap as a

function of S4.

The statistics summarized in Figure 4 include 17 days

for the GPStation-6 (collected between Nov 2012 and

Jan 2013) and 46 days for the PolaRxS Pro (collected

between Sep and Oct 2014). For both receivers the

probability of interrupted tracking on L1 increases as

a function of S4, but this probability was significantly

larger for the NovAtel than for the Septentrio.

Figure 6 summarizes the statistics of L2C tracking

for the PolaRxS Pro. Unfortunately, the GPStation-6

did not track L2C signals (presumably due to a

receiver configuration issue), so we cannot compare

results between the two receivers. The plots on the

left are organized in terms of the S4 index computed

for the L2C signal, while the plots on the right are

organized in terms of the S4 index computed for the

L1 signal. We fully expected a higher probability of

interrupted tracking of L2 for the same level of

ionospheric disturbance, because scintillations are

generally more intense on lower frequency carriers

due to the well-known frequency dependence of

scintillation. Comparing the plots on the right side of

Figure 6 with those on the right side of Figure 5

shows this expectation is clearly met (the probability

of a data gap is at least 5 times larger on L2C than

L1, for a given S4 value on L1. Nevertheless, we

anticipated a similar probability of interrupted

tracking on L1 as L2C when parameterized in terms

of their own signal fluctuations (i.e. the S4 on the

same carrier). This turned out to not be the case.

Comparing the left hand plots in Figure 6 with the

right hand plots in Figure 5, we see that the

probability of interrupted tracking is larger on the

lower frequency carrier, even when viewed in terms

of its own S4.

Figure 7 summarizes the statistics of L5 tracking for

the GPStation-6. Unfortunately, the GPStation-6 did

not track L5 signals (presumably due to a receiver

configuration issue), so we cannot compare results

between the two receivers. The plots on the left are

organized in terms of the S4 index computed for the

L5 signal, while the plots on the right are organized

in terms of the S4 index computed for the L1 signal.

This receiver experienced a higher probability of a

data gap on L5 than L1, for a given S4 value on L1.

Similarly, the probability of interrupted tracking is

larger on the L5 than L1, even when viewed in terms

of its own S4. More precisely stated, the probability

of a data gap on L5 (when viewed as a function of S4

on L5) was higher than the probability of a data gap

on L1 (when viewed as a function of S4 on L1).

We found the observation that the probability of

interrupted tracking was larger on L2C and L5 than

on L1, even when viewed in terms of their own S4

values to unexpected. This seems to suggest that L2C

and L5 tracking may be less robust to scintillation

effects than L1 tracking, despite the availability of

new codes on these GPS modernization signals and

the advanced algorithms that can be employed to

track them.


Figure 6. L2 tracking performance for the PolaRxS Pro as a function of S4 on L2 (left) and L1 (right).

Figure 7. L5 tracking performance for the GPStation-6 as a function of S4 on L5 (left) and L1 (right).


CONCLUSIONS

As part of our ongoing study to assess GNSS signal

tracking performance and navigation accuracy during

scintillation we evaluated the probability of

scintillation-induced loss of code lock on L1, L2C,

and L5 as encountered by two widely used GNSS

scintillation monitors, the NovAtel GPStation-6 and

Septentrio PolaRxS Pro. The approach we have taken

is to count the number of scintillation-induced gaps

in the high rate (50 Hz) receiver-reported signal

amplitudes. Next, we bin these data gaps as a

function of the scintillation index S4. The ratio of the

number of missing samples to the total number of

samples for a given S4 yields the probability of

interrupted code tracking as a function of S4. Only

high elevation observations are included in the

statistics to exclude fluctuations caused by multipath

reflections from terrestrial objects.

We found the probability of scintillation-induced L1

tracking interruption appears to be significantly

larger for the GPStation-6 than the PolaRxS Pro. We

also found that both receivers experienced a higher

probability of losing lock on the lower frequency

carriers, as one might expect due to the well-known

frequency dependence of scintillation. Perhaps less

expected, however, was the observation that that the

probability of interrupted tracking was larger on the

lower frequency carriers than on L1, even when

parameterized in terms of the S4 for their own

carriers. This seems to suggest that L2C and L5

tracking may be less robust to scintillation effects

than L1 tracking, despite the availability of new

codes on these GPS modernization signals and the

advanced algorithms that can be employed to track

them.

This study has several inherent limitations. Firstly,

the probabilities of losing lock we report are specific

to the particular GNSS receiver model tested. Due to

problems with data from one of our receivers we

were unable to compare the performance of the

GPStation-6 than the PolaRxS Pro during the same

time period. Also, there are many external factors

which could affect our results. Differences in the

receiver installation such as antenna types, cables,

and placement are a few possible hindrances to a fair

comparison of receiver tracking performance during

scintillation. Furthermore, the number of samples

included in this study is limited (particularly for L2C

and L5). Back in 2012, 10 satellites transmitted the

L2C signal while only 3 transmitted L5. Currently,

16 satellites transmit L2C and 8 transmit L5. We

hope to expand this study to include more recent data

in the future.

AKNOWLEDGEMENTS

This research was supported by Boston College

Cooperative Agreement FAA 11-G-006, sponsored

by Deane Bunce. The authors are indebted to Eurico

de Paula and João Francisco Galera Monico for

providing GNSS data from São José dos Campos.

REFERENCES

Aarons, J. (1982), Global morphology of ionospheric

scintillations, Proc. IEEE, 70, 360–378,

doi:10.1109/PROC.1982.12314.

Aarons, J. (1993), The longitudinal morphology of

equatorial F-layer irregularities relevant to their

occurrence, Space Sci. Rev. 63, 209-243.

Bougard, B., Sleewaegen, J-M., Spogli, L., Veettil,

Sreeja Vadakke, Monico, J.F. Galera (2011),

CIGALA: Challenging the Solar Maximum in Brazil

with PolaRxS, Proceedings of the 24th International

Technical Meeting of The Satellite Division of the

Institute of Navigation (ION GNSS 2011), Portland,

OR, September 2011, pp. 2572-2579.

Carrano, C. S., and Groves, K.M. (2010), Temporal

Decorrelation of GPS Satellite Signals due to

Multiple Scattering from Ionospheric Irregularities,

Proceedings of the 23rd International Technical

Meeting of The Satellite Division of the Institute of

Navigation (ION GNSS 2010), Portland, OR,

September 2010, pp. 361-374.

Carrano, C. S., Groves, K. M., McNeil, W. J.,

Doherty, P. H. (2012), Scintillation Characteristics

Across the GPS Frequency Band, Proceedings of the

25th International Technical Meeting of The Satellite

Division of the Institute of Navigation (ION GNSS

2012), Nashville, TN, September 2012, pp. 1972-

1989.

Carrano, C. S., K. Groves and J. Griffin (2005),

Empirical Characterization and Modeling of GPS

Positioning Errors Due to Ionospheric Scintillation,

Proceedings of the Ionospheric Effects Symposium,

Alexandria, VA, May 3-5, 2005

Carroll, M., Y. Morton, and E. Vinande (2014),

Triple Frequency GPS Signal Tracking During

Strong Ionospheric Scintillations over Ascension

Island, Position, Location and Navigation

Symposium - PLANS 2014, 2014 IEEE/ION, 5-8

May 2014, Monterey, CA.


Hlubek, N., J. Berdermann, V. Wilken, S. Gewies, N.

Jakowski, M. Wassaie, and B. Damtie (2014),

Scintillations of the GPS, GLONASS, and Galileo

signals at equatorial latitude, J. Space Weather Space

Clim. 4, A22.

Falletti E, Pini M, Presti L (2010), Are C/N0

algorithms equivalent in all situations, GNSS

Solutions, Inside GNSS, 5 (1), p. 20-27.

Peng, S., Y. Morton, W. Pelgrum, F. van Graas

(2011), High Latitude Ionosphere Scintillations at

GPS L5 Band, 24th International Technical Meeting

of the Satellite Division of the Institute of

Navigation, Portland, OR, September 19-23, 2011.

Shanmugam, S., J. Jones, and A. MacAulay, A.J.

Van Dierendonck (2012), Evolution to Modernized

GNSS Ionospheric Scintillation and TEC Monitoring,

IEEE/ION PLANS 2012 – April 24-26, Myrtle

Beach, SC.

Spoglia, L., Alfonsia, L., Romanoa, V., De

Franceschia, G.,Franciscob, G.,Shimabukurob, M.,

Bougarde, B., and Aquinod, M.(2013),Assessing the

GNSS scintillation climate over Brazil under

increasing solar activity, Journal of Atmospheric and

Solar-Terrestrial Physics Volumes 105–106,

December 2013, pp 199–206.

Van Dierendonck, A. J., J. Klobuchar, and Q. Hua

(1993), Ionospheric scintillation monitoring using

commercial single frequency C/A code receivers, in

ION GPS-93 Proceedings: Sixth International

Technical Meeting of the Satellite Division of the

Institute of Navigation, pp. 1333–1342, Inst. of

Navig., Salt Lake City, Utah.

A Statistical Analysis of GPS L1, L2, and L5 Tracking ... · PDF fileA Statistical Analysis of GPS L1, L2, and L5 ... Institute for Scientific Research, Boston College, ... Brazil,

Documents