A Performance Analysis of Future Global Navigation Satellite Systems

Paper Menu >>

Journal Menu >>

Journal of Global Positioning Systems (2004)

Vol. 3, No. 1-2: 232-241

A Performance Analysis of Future Global Navigation Satellite Systems

Cedric Seynat1, Allison Kealy2, Kefei Zhang3

1GPSat Systems Australia, Suite 1/22 Aberdeen Road, McLeod, Victoria,

email: cedric.seynat@gpsatsys.com.au, Tel: +61 (0)3 9455 0041 Fax: +61 (0)3 9455 0042

2Department of Geomatics, The University Of Melbourne, Victoria, Australia

email: akealy@unimelb.edu.au, Tel: +61 (0)3 8344 6804 Fax: + 61 (0)3 9347 2916

3School of Mathematical and Geospatial Sciences, RMIT University, Victoria, Australia

email: kefei.zhang@rmit.edu.au, Tel: +61 (0)3 9925 3272

Received: 15 Nov 2004 / Accepted: 3 Feb 2005

Abstract. For an increasing number of applications, the

performance characteristics of current generation Global

Navigation Satellite Systems (GNSS) cannot meet full

availability, accuracy, reliability, integrity and

vulnerability requirements. It is anticipated however that

around 2010 the next generation of GNSS will offer

around one hundred satellites for positioning and

navigation. This includes constellations from the US

modernised Global Positioning System, the Russian

Glonass, the European Galileo, the Japanese Quasi-

Zenith Satellite System and the Chinese Beidou. It is

predicted that the performance characteristics of GNSS

will be significantly improved. To maximise the potential

utility offered by this integrated infrastructure, this paper

presents an approach adopted in Australia to quantify the

performance improvements that will be available in the

future. It presents the design of a GNSS simulation

toolkit developed in Australia and the performance

expectations of future GNSS for a number of important

applications within the Asia Pacific region. In quantifying

the improvement in performance realised by combined

systems, this paper proposes a practical approach to

facilitate the development of innovative applications

based on future GNSS.

Key words: GPS, Galileo, GNSS, Simulation

1 Introduction

Currently, there are only two satellite navigation systems

in operation, the Global Positioning System (GPS), and

the Russian equivalent Glonass. The GPS signal is free

but its availability is not guaranteed and currently most

users are prepared to accept this risk. However, as

satellite navigation becomes a vital technology across a

number of critical industrial sectors, the prospect of, for

example, a nation’s transport infrastructure becoming

dependent on this technology is a strategic risk that most

industrial countries are not willing to accept. This

argument initiated the Galileo programme in Europe.

Galileo is a Global Navigation Satellite System (GNSS)

and 30 satellites orbiting the Earth at an altitude of

23,616km (three spares) will transmit a navigation signal

that can be received almost anywhere, and from which a

receiver can determine its position and time. Unlike GPS,

Galileo will also offer a guarantee of service to users who

are willing to pay for it (e.g. commercial service – CS,

and Public Regulated Service PRS) in addition to a free

signal similar to that of GPS (Open Service - OS and

Safety of Life service - SoL). Galileo will be available to

the public in 2010 (European Commission, 2003).

Despite many technical differences between these three

GNSS systems, the commonality of the carrier

frequencies they use creates the potential for the future

development of an interoperable GNSS receiver, as

illustrated in Table 1, which compares the services

available and associated signals both now and at around

2015. Aside from these three core GNSS infrastructures,

a number of additional space-based navigation systems

are also under development through various national

programmes. Japan is currently developing the Quasi-

Zenith Satellite System (QZSS), with three satellites

placed in a special orbit that maximises coverage over

Japan. The QZSS will complement other existing GNSS

systems over Japan, but at the same time, these satellites

will also be available over Australia and the South East

Asian region (Petrovski 2003). In China, the Beidou

Seynat et al.: A Performance Analysis of Future Global Navigation Satellite Systems 233

navigation system is also being developed. Current

Beidou satellite navigation and positioning system

consists two geosynchronous satellites based on the

DFH-3 bus. There shall be four satellites, two operational

and two backups upon completion of the system (Chinese

Defence Today, 2004). In addition, many augmentations

to GNSS are either under development, or have been

commissioned, or are already in use: Space Based

Augmentation Systems (SBAS) have been built (or are

being deployed) by the US, Europe, India, Japan, and

China. Ground-Based Augmentation Systems (GBAS)

offer tremendous performance benefits to the aviation

sector and have led to the development of the American

Wide Area Augmentation System (WAAS), Local Area

Augmentation Systems (LAAS) and the Australian

Ground-Based Regional Augmentation System (GRAS).

This brief overview of current and future navigation

infrastructures illustrates the huge potential that exists for

future navigation and positioning applications. The vast

majority of the world will be users of these existing

systems. The fundamental questions then are: “Which

system or systems should a country use?”; “How to

choose a combination of the systems?”; “What are the

benefits and respective merits of those systems?”. There

is no simple answer to these questions, as the best

solution will undoubtedly depend on the targeted

application, which has its own requirements in terms of

accuracy, reliability, robustness, cost and other

application-specific criteria. What can be provided,

however, is a means whereby parameters that describe

these performance requirements can be computed.

High-accuracy software simulations are a cost-effective

and precise approach of determining the performance

characteristics attainable from the future GNSS, and have

been recognised as an appropriate pre-development tool

for satellite navigation systems and applications in Japan

(Petrovski 2003) and Europe (Seynat 2003). The

technical benefits of this approach lie in the fact that the

simulations are reproducible and totally controlled, and

parameters can be changed individually if necessary for

an in-depth understanding of the underlying effects.

This paper introduces a simulation toolkit developed to

conduct a qualitative assessment of the performance

characteristics of the future GNSS infrastructure. The

design and development of the simulator architecture as

well as the models describing all influencing effects on

the performance of GNSS are presented. Finally,

representative results over Australia are demonstrated and

future developments are outline.

Table 1: Current and future GPS, Glonass and Galileo services and signals for 2004 and planned for 2015

GPS Glonass GALILEO

Services 2004 2015 2004 2015 2004 2015

Basic Positioning

(unencrypted)

SPS

9 L1 CA

SPS

9 L1 CA

9 L2C

9 L5

9 L1

9 L2

9 3rd Signal

9 L1

9 E5a

Integrity/safety

(unencrypted)

Integrity

message

SoL

9 L1

9 E5b

9 E5a

Commercial/value-added

(encrypted)

Security/military

(unencrypted)

PPS

9 L1 P(Y)

9 L2 P(Y)

PPS

9 L1 P(Y)

9 L2 P(Y)

9 L1 M

9 L2 M

9 L1

9 L2

9 L1

9 L2

9 Unknown

PRS

9 L1

9 E6

SPS-standard positioning service, PPS – precise position service, SP – standard precision, HP-high precision,

OS – open service, SoL – safety of life service, CS – commercial service, PRS – public regulated service

2 Simulation technology and methodology

Current and future users of navigation technologies need

to understand and quantify the performance they can

expect from the candidate systems, used individually or

in combinations. The GNSS Simulation Tool (GST)

developed in this research aims to provide a toolkit that

reproduces the performance behaviour of existing and

planned GNSS, in order to support the development of

234 Journal of Global Positioning Systems

next generation navigation applications. The main

objectives of the GST are:

9 To provide an accurate, independent tool capable of

analysing customised application scenarios, based on

realistic models of all effects relevant to the

performance of an application of the satellite

navigation technology;

9 To produce precise technical data on the performance

of navigation applications in the form of predefined or

custom-defined navigation scenarios;

9 To generate simulated navigation data as a real-world

navigation receiver would capture, in a format familiar

to application developers, in support of algorithm

development and testing; and

9 To allow external programs and file formats to connect

to the GST, in order to maximise the re-use of existing

expertise and data sharing.

To achieve these objectives, at the core of the GST is a

set of models relevant to the description of navigation

systems. These models are designed to be fully

configurable by the user. The GST also has a set of data

analysis tools and external file readers that either

initialise some of the GST model parameters or fully

replace a model, depending on the specific file used. In

addition, the GST can also be configured to use models

developed externally by third parties. These external

models can take the form of a Windows DLL or an

executable (“.exe”) file.

Figure 1: Functional diagram of the GST software architecture

3.1 Space Segment

Orbits: The GST computation of the satellites

coordinates is based on the standard Keplerian orbit

parameters. Satellite positions in the GST can also be

imported from actual data provided from real GPS

satellites, such as those provided on a daily basis by the

International GPS Service (IGS, 2004). The orbit files are

generated using the standard sp3 format, and the GST

contains a sp3 file reader. The values from the sp3 files

can be input in the GST in place of the computed

Keplerian coordinates. Alternatively, the GST can use

orbital parameters of the GPS constellation, as provided

in the YUMA format from the Internet (USCGNC, 2004).

Signal-in-space: The satellite signal is defined in the GST

by a set of signal characteristics:

9 The frequency f of the carrier;

9 The modulation scheme: Binary Phase Shift Keying

(BPSK) or Binary Offset Carrier (BOC); and

9 In case of the BOC modulation, the two integers n and

m, multiples of the base frequency 1.023MHz, defining

the sub-carrier frequency fs and the code rate fc,

Navigation message: The navigation message sent by the

satellite is not modelled in the GST. This means that the

satellite ephemeris and clock are not transmitted to the

receiver model and therefore not used in the position

computation of the receiver. However the ephemeris error

is represented in the GST as an error of eph

Figure 2: Functional diagram of the GST software architecture

3.2 Environment models

The term ‘environment’ refers here to the propagation

medium through which the satellite signals travel before

reaching the receiver antenna. The main effects

considered are the transmission delay due to ionospheric

effects, tropospheric effects and multipath.

Ionosphere: The transmission delay inferred by the

ionosphere is based on the knowledge of the total electron

content (TEC) along the transmission path of the signal

(Parkinson 1996). In order to obtain an accurate estimate

of the TEC, two approaches have been implemented in

the GST:

9 The first implementation uses a global ionospheric

model, NeQuick (Leitinger 1996), as an external model

plugged-in the GST architecture.

9 The second implementation uses measured data of the

vertical Total Electron Content (vTEC) regularly

available from public internet sources and published in

the Ionospheric Map Exchange (IONEX) format.

Models

and

algorithms

Inputs

Files readers

Data analysis

External models

and algorithms

Output and

visualisation

Seynat et al.: A Performance Analysis of Future Global Navigation Satellite Systems 235

The two implementations described above for modelling

of the ionospheric effect on the signal transmission are

complementary rather than exclusive. While TEC maps

provided by the IONEX files are more representative of

the actual situation that one wants to simulate with the

GST on a specific date, time, and location, the NeQuick

model, is more generic and can be used to simulate a data

and time set in the future.

Troposphere: The signal transmission delay due to

tropospheric effects is modelled using the refraction

index of the troposphere (Di Giovanni 1990). The

Hopfield model is used to estimate the refraction index

(Hofmann 1998). The model is dependent on the

estimates of local temperature, atmospheric pressure and

water vapour partial pressure. In the GST these values

can be set manually or alternatively, the GST can read

meteorological data collected at ground reference stations

and output in the Receiver Independent Exchange

(RINEX) format (Gurtner 2002).

Multipath: Multipath is highly dependent on the local

environment surrounding the navigation receiver.

Designed to be a generic tool, the GST models the

multipath effect as its end effect on the measured range.

Multipath causes a range measurement error, which can

be isolated from other error sources in actual receiver

measurements. The model used in the GST is an

empirical model based on observations of time series of

multipath range error (Parkinson 1996). The range error

caused by multipath, mult

, is modelled as follows:

()

()()cos ()

multmultmult c

tb antkt

δη

=+ (1)

where bmult is the multipath bias error, amult is the

amplitude of the multipath error,

is the satellite

elevation angle at the receiver location, k is a factor

required to adjust the elevation dependence, and nc(t) is

correlated Gaussian noise.

The model in Equation (1) is appropriate to represent the

range error caused by the multipath for several reasons.

First, the elevation dependence is modelled in such a way

that satellite at a low elevation causes higher multipath

errors, as it is observed from actual measurements. The

factor “k” allows adjusting the peak of this elevation

dependence. Secondly, the use of correlated Gaussian

noise also represents effects seen in actual observations.

Typical correlation times observed in multipath effects

are of the order of a few minutes. To simplify the tuning

of this model, the GST proposes several sets of default

values corresponding to low, medium, and strong

multipath environments. Expert users are also allowed to

change parameters separately if they wish.

3.3 User Segment

Receiver coordinates: Receiver coordinates are input in

the GST as a latitude, longitude, and height in WGS-84.

Masking angle: The GST allows the definition of a

customised masking profile to represent the specific

situation at a particular location.

Receiver carrier-to-noise ratio: The total signal carrier-

to-noise ratio is a measure of the signal quality and

influences the tracking error. It is modelled as the sum of

the signal-to-noise ratio where no jamming is present,

plus the jammer-to noise ratio.

Receiver tracking error: The modelling of the accuracy

of the range measurement in the GST assumes a Non-

Coherent Early-Late Processing. Two models are

implemented, one for the GPS BPSK modulation scheme

(Kaplan 1996) and one for the Galileo BOC scheme (Betz

2000).

Receiver position: The receiver model computes an

estimation of its position based on measured range

between itself and each satellite in view. In the current

implementation of the GST, the measured range is

modelled as the true satellite-to-receiver range plus the

sum of the errors described in the previous sections. Then

the position is estimated using a weighted least square

algorithm.

User Equivalent Range Error: The models described in

the previous sections account precisely for the different

error sources affecting the navigation solution. An

alternative, less accurate, modelling of these errors is

commonly used when simulations need to be run for long

simulated periods and over large geographical areas. In

this alternative method, the errors are grouped in a User

Equivalent Range Error (UERE). In the GST the UERE

for each satellite is a function of its elevation. Published

UERE values for GPS and Galileo were used (Shaw

2000, Benedicto 2000).

3.4 GST predefined figures of merits

To describe the output of the GST a number of standard

performance parameters are computed.

9 Availability of navigation solution: The receiver can

make an estimate of its position and clock bias when at

least 4 satellites are in view. The instantaneous

availability of the navigation solution ()

nav

is a flag

set to 1 if at least four satellites are visible at a time t,

and set to 0 otherwise: The availability of the

navigation solution in a time window is the percentage

of the time, in the time window, when () 1

nav

236 Journal of Global Positioning Systems

9 Continuity of navigation solution: The continuity of the

navigation solution is a quantitative estimate whether

the navigation solution can be computed without

interruption. The instantaneous continuity of the

navigation solution,

()

nav

, is a flag set to 1 if the

navigation solution is available at the time t and time t-

DT, and set to 0 otherwise. The continuity of the

navigation solution in a time window is the percentage

of the time in the time window when () 1

nav

ct=.

9 Availability of Accuracy: The availability of the

accuracy expresses whether the position estimate made

by the receiver is accurate enough to be used for a

particular application. Different applications have

different accuracy requirements, and if the position

computed by the receiver is not accurate enough for a

given application to be successfully carried out, then

satellite navigation is not appropriate. The

instantaneous availability of accuracy, ()

acc

, is a flag

set to 1 if the positioning error of the receiver is less

than a user-defined threshold, and set to 0 otherwise.

The availability of the accuracy in a time window is the

percentage of the time when ()1

acc

at=.

9 Continuity of accuracy: The continuity of the accuracy

is equivalent to the continuity of navigation solution

defined previously, using the availability of accuracy to

estimate continuity.

9 Dilution of precision: The Dilution Of Precision (DOP)

is a well-known indicator of the geometry of the

satellites in view from a receiver.

9 Position error: The instantaneous position error is the

scalar difference between the receiver true position and

the position estimated by the single point positioning

algorithm of the receiver model.

3.5 Model Validation

Each of the models above has been carefully tested

individually. Also, the GST results have been validated

by independent sources. It is beyond the scope of this

paper to present a detailed validation report of the

models. Validation has been carried out in the following

areas:

9 Receiver and satellite geometry: coordinates, line-of-

sight vectors, and DOP values were compared against

those obtained from a separate software tool (Satellite

ToolKit)

9 Receiver RAIM availability figures were compared

against a service volume simulator developed in

Europe (Loizou 2002).

9 World and Europe performance maps published in

(Lachapelle 2002) and (Leonard 2003) were

successfully reproduced with the GST

9 The simulated receiver pseudorange and carrier

measurements were input in the GPS/Glonass/SBAS

data Toolkit Teqc (UNAVCO 2004) and successfully

tested for quality.

Other comparisons have been conducted and further

investigation is currently being conducted by the authors.

4 Simulation Results

This section presents a sample of the typical simulation

results obtained from the GST spanning across the entire

suite of the models described earlier. The aim of this

section is to illustrate the capabilities of the GST in

simulations focused on Australia. The results presented

here cover a wide range of topics and this will be the first

investigation undertaken in Australia.

4.1 Availability of Navigation Solution over Australia

The global availability of the navigation solution over

Australia is dependent upon the number of satellites in

view and the mask angle. Simulations were run at a 40°

mask angle for the Galileo constellation only, the GPS

constellation only, and for the combined GPS+Galileo

constellation. Simulations of 24 hours were run.

At a 40° mask angle, both the GPS and Galileo

constellation when used alone offers reduced availability

of the navigation solution. The GPS constellation is

unusable most of the time (about 20% availability on

average), and the Galileo constellation offers about 60%

availability almost everywhere except in the

northernmost region of the country.

The clear latitude dependence shown in Figure 3(b) is due

to the spherical symmetry of the Galileo constellation

with respect to the Earth’s centre. The advantage of using

a combined GPS and Galileo constellation is clearly

shown on Figure 3, as the availability of the navigation

solution remains 100% globally.

Seynat et al.: A Performance Analysis of Future Global Navigation Satellite Systems 237

(a) GPS

(b) Galileo

Figure 3 - Availability of Navigation Solution over Australia for a 40°

mask angle

(a) Galileo

(b) GPS

Figure 4 - 95% Percentile accuracy for a dual-frequency (a) L1/E5a

Galileo receiver, (b) L1/L5 GPS receiver and (c) combined Galileo/GPS

receiver at the same 2 frequencies

238 Journal of Global Positioning Systems

4.2 Positioning Accuracy

The positioning accuracy presented in this section was

obtained by using the GST in Service Volume mode. The

UERE budget used for this purpose was a dual frequency

L1/E5a Galileo receiver and dual frequency L1/L5 GPS

receiver. Of course such receivers do not exist yet but

their expected performance was presented in (Shaw, 2000;

Benedicto, 2000) and are used here. A constant mask

angle of 20° was used, for a simulation time of 24 hours.

Figure 4 shows that the 95% accuracy to be expected

from Galileo or GPS dual-frequency receivers is of the

same order of magnitude (about 6m across the country).

The combined performance of Galileo and GPS is

expected to be of about 2-3m, which is a significant

improvement compared to each individual system

The effect of using a combined GPS/Galileo receiver is

also to remove geographical disparity across the country.

The time traces of positioning accuracy (not displayed

here) are also more consistent, with less fluctuation. This

effect of combined use is as important as the absolute

gain in accuracy itself. Users using Galileo/GPS receivers

can expect a consistent performance of their system,

which may reduce the need for costly augmentations in

areas and at times of the day where individual systems

fail to deliver the required performance level.

4.3 Impact of QZSS on navigation performance in

Australia

The QZSS has the primary objective of augmenting GPS

over Japan so that satellite availability in dense urban

areas like Tokyo remains high. The orbit of the QZSS

satellites is such that there will always be at least one, and

often 2 satellites in view and at high elevation from

receivers located in Japan. However, the QZSS satellites

also pass above Australia, and for that reason they offer

the potential of augmenting GPS in Australia as well.

This potential is assessed here with the GST.

The design of the QZSS constellation orbits is presented

by Petrovski (2003), with several options being

considered including an “8-shape” orbit used in

subsequent GST simulation shown here. The QZSS

satellites will pass above or near Australia. It is therefore

possible to envisage that those satellites can be used in

Australia, in the same way as they will be used in Japan,

i.e. as an augmentation to GPS. The number of visible

QZSS satellites and their elevation is presented here for

four Australian state capitals, namely Melbourne,

Sydney, Darwin and Perth. These locations have been

chosen because they are widespread across the country

and because the benefits of using QZSS in urban areas

will be illustrated later, particularly in low visibility

situations that can occur in Melbourne or Sydney.

Figure 6 indicates that QZSS satellites in an “8-shape”

orbit will be well visible from Australia. In Melbourne

and Sydney, at least one satellite will be visible above

60° at all times. In Darwin, all 3 satellites will be visible

at all times from locations with mask angles lower than

25. The figures from Perth are also excellent.

Figure 5 shows that QZSS is particularly relevant to

Australian GNSS users. In areas of reduced visibility, the

availability of a single additional satellite in all times can

make a difference between getting a position fix or not.

At least one QZSS satellite will be almost always in view

at any time, even in urban canyons. With this preliminary

result based on geometry, it appears that QZSS usage in

Australia should be investigated further, at both technical

and political levels. To illustrate the navigation

improvement resulted from the combined use of GPS and

QZSS in Australia, the simulation results in an urban area

is presented in Figure 6, for the location of Melbourne,

and a mask profile representing an urban canyon.

Figure 6 shows the availability and continuity of the

accuracy in an urban environment at the location of

Melbourne for the GPS only and combined GPS/QZSS

cases. The simulation uses a single frequency receiver

and the plot displayed here is relevant to mass market

applications, such as personal mobility. The gain in

availability and continuity of accuracy is clear from these

plots, although for an accuracy threshold of 12m, 100%

availability and continuity is not achieved at all times

even with combined QZSS+GPS receivers. Additional

systems, such as the addition of Galileo or pseudolites,

may be required depending on the application considered.

Investigation into pseudolite networks in urban

environments is a potential field of study that can be

made with the GST.

Overall, the results presented in the previous three

sections show the significant improvement in navigation

performance to be expected from a receiver capable of

tracking several independent systems. More than in

positioning accuracy, the improvement is in the

consistency of the performance and therefore in the

reliability of the application from the end-user

perspective. This is an important consideration from a

marketing perspective, as users are more likely to adopt

the new technology if its reliability is certified

4.4 Raw data generation for post-processing by third

party software

Based on its accurate models and its end-to-end

capability, the GST can be used to generate simulated

measurement as they would be received by a real

receiver. The GST uses the widely accepted RINEX 2.1

format to output simulated data. The GST generates the

RINEX observation and navigation files. Part of the

Seynat et al.: A Performance Analysis of Future Global Navigation Satellite Systems 239

validation exercise outlined in Section 3.5 was to produce

RINEX files and input them in the TEQC software for

quality check. More generally, the availability of

simulated Galileo and GPS data opens the possibility for

research organisations to start testing algorithms and

applications early, in readiness for the future availability

of the hardware. This approach saves development and

testing effort and can be achieved using the GST. Such

capabilities will be illustrated in further publications.

(a) Melbourne

(b)Sydney

(c)Darwin

Figure 5 “8-shape” QZSS satellite elevation (left) and number of visible satellites (right) at Melbourne, Sydney and Darwin in Australia

240 Journal of Global Positioning Systems

Figure 6 Availability of Accuracy and Continuity of Accuracy from a receiver in an urban area at the location of Melbourne, obtained from an

accuracy threshold of 12m

5 Conclusions

This paper presents the GNSS Simulation Tool, a

software tool of navigation systems that aims to provide a

means for research organisations and industry to develop

navigation applications using the current and emerging

technologies.

The reason to build such a tool today originates from the

fact that navigation technology, and especially satellite

navigation, is now at a major crossroad: new satellite

systems are being built in Europe, Japan, India, China

and the United States. Current systems are being

maintained, augmented and rapidly improved, in Russia

and the United States. Adding to the advent of these new

technologies is the booming need in today’s society for

positioning information. It is the dual growth of the

technology and the market that makes the future of

navigation multidisciplinary and challenging. The

current GST itself cannot handle fully the complexity

and variety of the challenges to come in the near

future. However, it provides a low-cost, flexible tool to

provide specific answers to a number of performance