34

CHAPTER 1

INTRODUCTION

1.1 Back ground

Navigation is the process of identifying, directing and controlling

the movement of an object/person from one place to another. Satellite

based navigation systems such as the Global Positioning System (GPS)

are being implemented now days all over the world due to their global

coverage and operational ease. At present GPS is the only available

fully operational Global navigation satellite system (GNSS) [1].

However the required accuracy, availability and integrity, cannot be

obtained with stand-alone GPS, which are very important for precision

approach applications in both civilian and defense sectors.

The desired positional accuracy over a pre-defined area can be

obtained by augmenting the GPS. In recent times, several countries

are developing their own Satellite Based Augmentation Systems

(SBAS), for their increased safety and security [2-4]. The different

SBAS systems are the Wide Area Augmentation System (WAAS),

operated by the United States Federal Aviation Administration (FAA)

[5], Russia’s GLONASS (Global Navigation Satellite System) [6-8], the

European Geostationary Navigation Overlay Service (EGNOS),

operated by the European Space Agency [9], the Wide Area GPS

Enhancement (WAGE), operated by the United States Department of

Defense for use by military and authorized receivers [10], the Multi-

functional Satellite Augmentation System (MSAS), operated by Japan's

35

Ministry of Land, Infrastructure and Transport (JCAB) [11], the Quasi-

Zenith Satellite System (QZSS), proposed by Japan [12]and China’s

regional Beidou navigation system[13].

In India also, Indian Space Research Organization (ISRO) and

Airport authority of India (AAI) are jointly developing “GAGAN”

(GPS Aided GEO Augmented Navigation) system over the Indian

air space and is expected to become operational by 2014. The basic

aim of GAGAN is to provide navigation for all phases of flight over the

Indian air space and in the neighboring area especially for strategic

defense applications [14].

The positional accuracy of GPS is predominantly affected by the

ionospheric time delay, which is a function of Total Electron content

(TEC) [1]. TEC varies greatly with the amount of radiation received

from the sun. Thus there is a diurnal (time of day) effect and seasonal

effect on the TEC. The activity of the sun is associated with

the sunspot cycle and increased radiation results in more sunspots

[15]. Further, the distribution of ionospheric plasma is affected by

the solar and magnetic disturbances like the occurrences of solar

flares and magnetic storms [16-21]. Major magnetic storms

originated from solar bursts can cause strong disturbances in the

geo space environment [22-25]. These storms are usually

associated with increased electron densities in the lower

ionosphere and result in simultaneous increase in absorption of

radio waves [26-27]. During ionospheric disturbance periods, the

positioning errors may exceed 50 meters, and can reach up to 150

36

meters under extreme solar activities, at mid day, and near the

horizon, which ultimately affect the accuracy of Global Navigation

Satellite Systems (GNSS) [28]. Hence the accurate specification and

prediction of ionospheric conditions for both quiet and disturbed

periods will aid in the efficient design and operation of GNSS.

Moreover, as India comes under equatorial and low latitude

regions, the prediction of TEC is especially difficult. The ionospheric

variations are more severe in the equatorial region, covering about

±200 dip around the magnetic equator. In this region, the ionospheric

behavior is highly volatile with large horizontal gradients and day-to-

day variability [29-30]. Hence, the primary task in developing an

ionospheric TEC prediction model for the Indian conditions will be to

take inputs from several fixed reference stations across the Indian

subcontinent by using dual frequency GPS Receivers and model the

data [31]. However, in India only two stations data is available in the

Scripps Orbit and Permanent Array Center (SOPAC) data archive of

the International GNSS Service (IGS) [32]. Further, as the already

available Global empirical electron density prediction models such as

the “International Reference Ionosphere” (IRI) [33] have some

limitations for application to the Indian latitudes as less data from

the Indian subcontinent is used in the model development.

Hence the analysis of TEC variations over some of the low latitude

stations around the globe is taken up in this thesis to develop a region

specific TEC model. Ten stations are considered from the northern

hemisphere since India is situated on the same side of the globe.

37

Further, out of these stations, six were considered from eastern side

and the remaining four are from western side with reference to the

Greenwich line. The stations are Singapore (1.350N, 103.680E), Medan

(3.620N, 98.710E), Managua (12.150N, 86.250W), Bangalore (13.030N,

77.510E), Guatemala (14.590N, 0.530W), Quezon (14.640N, 121.080E),

Dakar (14.680N, 17.470W), Hyderabad (17.420N, 78.550E),

Christiansted (17.760N, 64.580W) and Kunming (25.030N, 102.800E).

Afterwards, the TEC data is also used to develop a region specific

prediction model for the low latitude stations using Neural Networks.

1.2 Objectives and Contributions of thesis

The main objective of this thesis is to analyze the daily, monthly,

seasonal and storm time variations of TEC over a few low latitude

stations around the globe and to develop an empirical region-specific

ionospheric TEC prediction model using neural networks. As part of

this objective, data from ten low latitude stations is taken up and the

following works are carried out:

(i) The diurnal, monthly and seasonal variations of TEC and the

Cumulative Probability of Range Delays are analyzed during

the year 2003.

(ii) The TEC values during the major geomagnetic storm days

from January, 2003 to July, 2012 are analyzed. Further, the

TEC values recorded during the severe geomagnetic storm in

October 2003 are compared with different prediction model

results to verify the validity of these models for low latitude

conditions.

38

(iii) A neural network based TEC model has been developed

using the back propagation algorithm.

1.3 Applications of thesis

Ionospheric time delay statistics are necessary for evaluating the

performance and reliability of the communication and navigation

systems. The statistical analysis of ionospheric time delay for ten low

latitude stations around the globe over a complete year 2003, reported

in this thesis can be extended to other stations for predicting the

ionospheric variability in multiple directions, around the reference or

user stations.

The analysis of diurnal, monthly, seasonal and storm time

variations of TEC over ten low latitude stations reported in this thesis

can be extended to other intended stations so as to obtain a complete

idea about the ionospheric conditions prevalent over that particular

region.

The methodology used to obtain the neural network based TEC

prediction model for a group of low latitude stations can be extended

further to other stations/regions in order to develop a more complete

model.

1.4 Motivation of thesis

Several parameters such as the ionospheric time delay,

tropospheric delay, multipath error, ephemeris error and the errors in

the receiver system will influence the positional accuracy of stand-

alone GPS [1]. Of all these errors, ionospheric time delay is the most

predominant which is a function of TEC. Therefore, the quality of local

39

TEC information will be one of the key issues to develop region-specific

ionospheric models accurately. Several researchers in India have

attempted to model the ionospheric parameters, but till now no

significant work based on GPS data is reported on model development

to suit the Indian conditions.

Ionospheric time delay statistics (daily, monthly, seasonal and

storm-time variations of TEC) are necessary in order to provide a

better region-specific ionospheric model for the Indian subcontinent

spread over 6°N-38°N latitude range [34]. Further, as India is

developing its own navigation system, “GAGAN”, a suitable

ionospheric TEC model is necessary for precision approach

applications. This aspect motivated us to consider the present task

of developing a region specific ionospheric TEC model based on GPS

observations at a few low latitude stations.

1.5 Literature survey The GPS satellite signals travel through the ionosphere on their

way to the GPS receivers. The radio signal experiences an Ionospheric

time delay due to the dispersive nature of the ionosphere. Prior

knowledge of the GPS fundamentals and ionospheric effects on GPS

signals is necessary to complete the objectives of this thesis.

The wave propagation effects on satellite communication systems

are presented by different authors such as Ippolito,L.J.,Jr. [35]

Brussard and Rogers [36] and Timothy Pratt, Charles W.Bostian and

Jermy E. Allnutt [37]. Ionospheric effects on Earth-Space propagation

are in general reported in 1983 [38] and on GPS in 1991[15] by John

40

A.Klobuchar. Some satellite-to-ground propagation problems in the

UHF and L- bands caused by the Earth’s ionosphere are reported by

Kenneth Davies and Earnest K. Smith in 2002 [39]. An overview of

propagation problems in Satellite Navigation is reported by A.

Hornbostel in 2007 [40].

GPS principles and signal structure are reported by Ananda M [41],

Parkinson [1], Kaplan [42], Misra and Enge [43], Hoffman-

Wellenhof,B., Lichtenegger,H. and Collins,J. [44] and Ahmed El-

Rabbany [28]. Various errors that limit the positional accuracy of GPS

are dealt by Leick [45] and Langley [46]. The positional accuracy of

GPS is predominantly affected by the ionospheric time delay, which is

a function of Total Electron Content (TEC) [1]. TEC varies greatly with

the amount of radiation received from the sun. Thus there is

a diurnal (time of day) and seasonal effects on the TEC. Further, the

distribution of ionospheric plasma is affected by the solar and

magnetic disturbances like the occurrences of so lar flares and

magnetic storms [16-21]. Major magnetic storms originated from

solar bursts can cause strong disturbances in the geo space

environment [22-25].

Geomagnetic storm effects at heights of about 0-100 km in mid

latitudes, particularly Europe, is reported by Lastovicka,J. in 1996

[23]. The results of an investigation on the sequence of events from

the sun to the Earth that ultimately led to the major geomagnetic

storms is reported by Zhang,J., Richardson,G., Webb,D.F.,

Gopalaswami,N., Huttunen,E., Kasper,J.C., Nitta,N.V., Poomvises,W.

41

Thompson,B.J., Wu C.C., Yashiro,S. and Zhukov, A.N. in 2007 [22].

The research on historical geomagnetic storms is reported by

Lakhina,G.S., Alex,S., Tsurutani,B.T. and Gonzalez,W.D. in 2005 [47].

The solar-terrestrial events of October 2003 Halloween storm is

reported by Gopalswami,N., Barbieri,L., Cliver,E.W., Lu,G., Plunkett,

S.P. and Skoug,R.M. in 2005 [48]. The extreme Halloween 2003 solar

flares (and Bastille Day, 2000 Flare), Interplanetary Coronal Mass

Ejections (ICMEs), and resultant extreme ionospheric effects: A review

is reported by B.T.Tsurutani, A.J.Mannucci, B.Iijima, F.L.Guarnieri,

W.D.Gonzalez, D.L.Judge, P.Gangopadhyay and J.Pap in 2006 [49].

The global characteristics of the ionospheric storm and irregularities

as well as propagation of TEC disturbances during the strong

magnetic storm occurred in November 2004 is reported by XU Liang,

CHENG Guang Hui, XU Jisheng and LIU YongMin in 2008 [26].

Magnetic Storms and their influence on Navigation is reported by

R.Cop, S.Mihajlovic and LJ.R.Cander in 2008 [50].

The study of large geomagnetic storms of solar cycle 23 to

understand their solar, interplanetary and geospace conditions is

reported by N.Gopalswami in 2009 [51]. Geomagnetic storm effects on

GPS based navigation has been carried out by P. V. S. Rama Rao,

S.Gopi Krishna, J. Vara Prasad, S.N.V.S. Prasad, D.S.V.V.D. Prasad

and K. Niranjan in 2009 [52]. Study of ionospheric variability during

geomagnetic storms is reported by Rakhee Malik, Shivalika Sarkar,

Shweta Mukherjee and A.K. Gwal in 2010 [18]. Geomagnetic

observations and ionospheric response during storm on 14th April,

42

2006 for seven ionosonde stations located in the American sector is

reported by Bakare,N.O., Chukwuma,V.U. and Adekoya,B.J. in 2010

[53]. The relation between TEC variations and the magnetic index AP of

the geomagnetic storms for Oran city in Algeria for the period of the

month of January 2004 is reported by Seddik Boutiouta and Ahmed

Hafid Belbachir in 2006 [27].

Effect of major geomagnetic storms on TEC variations over

equatorial low latitude regions have been carried out by many

researchers. Ionospheric total electron content (TEC) studies with GPS

in the equatorial region is reported by Das Gupta, A., Paul, A and Das,

A. [29] in 2007. Local time dependent response of Indian equatorial

ionosphere to the moderate geomagnetic storms is reported by

S.Tulasi Ram, P.V.S Rama Rao, D.S.V.V.D Prasad, K.Niranjan,

R.Sridharan, C.V.Devasia and Sudha Ravindran in 2007 [54]. TEC

variations during low solar activity period (2005-2007) near the

equatorial ionospheric anomaly crest region in India have been

reported by Mala S.Bagiya , Joshi, H.P., Iyer, K.N., Aggarwal, M.,

Ravindran, S. and Pathan, B. M in 2009 [55]. Low latitude

geomagnetic response to the interplanetary conditions during very

intense magnetic storms is reported by Rawat,R., Alex,S.,

Lakhina,G.S. in 2009 [19]. Large enhancements in low latitude TEC

during 15th May, 2005 geomagnetic storm in Indian zone is carried

out by N.Dashora, S.Sharma, R.S.Dabas, S.Alex and R.Pandey in

2009 [56]. The effect of geomagnetic storm on GPS derived Total

Electron Content (TEC) at Varanasi, India is reported by Sanjay

43

Kumar and Singh, A.K. in 2010 [57]. Single and dual frequency GPS

receivers used in low latitude regions can suffer from rapid amplitude

and phase fluctuations known as Scintillations. These ionospheric

scintillation effects are reported by Datta-Barua,S., Dohetry,P.H. and

Delay,S.H.in 2003 [58].

Accurate prediction of TEC for both quiet and disturbed periods

will be useful in developing an ionospheric prediction model. There

are Global empirical electron density prediction models to estimate

the TEC. Various ionospheric propagation models and their limitations

are reported by different authors such as Somayajulu, Rush C.M. and

Cander Lj.R. [59-62]. Different prominent density models such as

Bent and International Reference Ionosphere (IRI)-2007 are

thoroughly understood. The Bent model [63] is an empirical worldwide

algorithm and estimates the electron density profile, the associated

delay and directional changes of a wave due to refraction. This model

accounts for up to 80% of the total ionospheric effect. Later Klobuchar

developed a time delay algorithm based on Bent’s electron density

model, which is more suitable for the mid latitude regions only [64].

The IRI is another most commonly used global empirical electron

density prediction model. The IRI is a joint project of the Committee

on Space Research (COSPAR) and the International Union of Radio

science (URSI). IRI is an empirical model specifying monthly

averages of electron density, ion composition, electron temperature

and ion temperature in the altitude range from 50 km to 1500 km.

Several steadily improved versions of the model have been released

44

with IRI-2007 being the latest version available for users to predict

the electron density of the ionosphere [33]. This model is

recommended by International Telecommunication Union, Radio

communication sector ITU-R (ITU 2004) as a suitable method for

TEC estimation.

IGS analysis centers including Jet Propulsion Laboratory (JPL),

Center for Orbit Determination in Europe (CODE), European Space

Operations Center (ESOC), Polytechnical University of Catalonia

(UPC) are also generating TEC estimations [65]. Estimation of vertical

TEC from GPS data has been reported by Arikan,F., Erol,C.B. and

Arican,O. [66] in 2003. A neural network approach for regional

Vertical Total Electron Content (VTEC) modeling has been carried out

by R.F.Leandro and M.C.Santos in 2006 using data sets collected by

the Brazilian GPS network(RMBC) [67]. Web Based Automated Total

Electron Content computation is reported by Orhan Ugurlu, Umut

sezen and Ali Ziya Alkar in 2007 [68]. Recently Prediction of

ionospheric total electron content using adaptive neural network with

in-situ learning algorithm is reported by Rajat Acharya, Bijoy Roy,

M.R. Sivaraman and Ashish Dasgupta in 2011 [69].

Many contributions are reported in the development of ionospheric

models for middle latitude regions. At the same time there is little

correspondence related to the low latitude ionospheric behavior. As

India comes under the low-latitude region, more care is to be taken

in developing the region specific ionospheric prediction models. A

numerical model for low latitude ionospheric TEC is reported by

45

Sethia,G., Chandra, H., Deshpande,M.R. and Rastogi, R.G. in 1978

[70]. Baruah. S and Bhuyan. P.K, reported a regional TEC model for

the Indian region in the year 2000 [71]. Development of Grid Based

Model for GAGAN is reported by Rajat acharya, Yadav,P.D., Vipin

Chandra Pant and Sivaraman,M.R. in 2005 [72]. Modeling of Indian

ionosphere using Minimum Mean Square Error (MMSE) technique

for GAGAN applications is reported by D.Venkata Ratnam and

A.D.Sarma in 2006 [73]. TEC derived from GPS network in India and

Comparison with models has been carried out by P.K. Bhuyan and

Rashmi Rekha Borah in the year 2007 [74].

In general, Ionospheric prediction models are either physically,

statistically or empirically – based [60]. Although several prediction

methods are available to generate and present the TEC, either

the temporal resolution is low or the estimations are based on

empirical data. This clearly demonstrates the necessity of developing

a new ionospheric TEC prediction model to suit the Indian conditions

i.e. GAGAN.

1.6 Global Positioning System (GPS)

The United States Department of Defense (DOD) decided to develop

a satellite navigation system named as Navigation Satellite Timing And

Ranging Global Positioning System (NAVSTAR GPS) commonly

known GPS in 1973[44]. In 1978, the first GPS satellite is launched

and the system became fully operational from 1995. The GPS system

consists of three segments namely space segment (SS), control

segment (CS), and user segment (US) as shown in Fig.1.1.

46

Fig.1.1 Basic system elements of the GPS

(i) Space Segment

In Space segment the satellite vehicles (SV’s) are positioned in six

earth-centered orbital planes at an altitude of 20,180 Kilo meters from

the earth surface. The orbital planes are equally separated by 600

above the equator with an inclination of approximately 550 relative to

the equator. The GPS satellites travel with a velocity of 3.9km/s and

hence will take 11 hours 58 minutes for one complete revolution

round the earth. The orbits are arranged so that at least six satellites

are always within line of sight from almost everywhere on Earth's

surface as shown in Fig.1.2 [75].

SPACE SEGMENT

(L1, L2) PSEUDO-RANGE DATA, CURRENT

EPHEMERIS CLOCK CORRECTIONS,

IONOSPHERIC DATA

CONTROL SEGMENT

4 SELECTED SATELLITES EACH WITH PRECISION TIME STANDARD PSEUDO-

RANDOM DATA

MONITOR STATIONS

HAWAII

ASCENSION ISLAND DIEGO GARCIA

KWAJALEIN COLORADO

SPRINGS CAPE

CANAVERAL

UPLOAD STATIONS

ASCENSION ISLAND

DIEGO GARCIA

KWAJALEIN

CAPE CANAVERAL

MASTER CONTROL STATION

COLORADO SPRINGS

RECEIVER ACCURATE POSITION VELOCITY

TIME

(L1, L2) PSEDO-RANGE DATA

EPHEMERIS CLOCK CORRECTIONS

IONOSPHERIC DATA

47

Fig.1.2 GPS satellite constellation.

(http://www.colorodo.edu/geography/gcraft/notes/gps/gif/orbits.gif)

The current GPS constellation consists of 31 Block II/IIA/IIR/IIR-M/IIF

satellites [76]. GPS constellation and individual satellite status is

updated every working day. Block-I satellites are referred to as the

original concept validation satellites developed by Rockwell

International and reflect various stages of system development. The

first Block-II satellite is launched in February 1989 and the most recent

Block-IIF satellite is launched on 04 October 2012 [77]. With more

number of satellite vehicles, the precision of GPS receiver calculations

can be improved by providing redundant measurements. With new

constellation around 8-10 satellites are visible from any point on the

ground at any given time.

The GPS satellites transmit two spread-spectrum pseudo-random

noise (PRN) signals. The signals consist of C/A (coarse acquisition)

code at 1.023 M Hz and P (precision) code at 10.23 M Hz bandwidth.

48

The two signals are transmitted at frequencies of 1575.42 M Hz (L1)

and 1227.60 M Hz (L2) respectively. Both are coherently derived from

highly stable on board atomic clocks. Both C/A and P–codes are

transmitted on the L1 frequency, whereas either C/A code or P–code

is transmitted on the L2 frequency depending on the ground

command [41]. The L3 signal (1381.05 M Hz) is a non–navigation

signal used for the nuclear detonation detection. L4 (1379.913 M Hz/

1841.4 M Hz) signal is useful to study the additional ionospheric

corrections and L5 (1176.45 M Hz) signal is reserved for safety-to-life

data/pilot signal [43].

GPS basically offers two types of services, namely Standard

Positioning Service (SPS) and Precise Positioning Service (PPS). SPS

is available for civilian use and broadcast at a single frequency. PPS

is meant for military use and broadcast using two frequencies. C/A

code is available for all users (SPS) where as P-code is available to

authorized users only (PPS). A navigation message comprising both

the ephemeris and clock parameters are also modulated on to the

PRN sequence on both L1 and L2 frequencies. Each satellite

transmits its identity number, the time and orbital ephemeris

correction, satellite health, clock errors, drift rates etc.

(ii) Control Segment

The control segment consists of a worldwide system of tracking

and monitoring stations. The 'Master Control Facility' is located at

Colorado Springs. The monitor stations collect information from the

GPS satellites and relay it to the Master Control Station (MCS). The

49

MCS uses this data to compute precise orbital models for the entire

GPS constellation. This information is then formatted into updated

navigation messages for each satellite. The navigation message is

uplinked three times daily for better system accuracy. The control

segment also maintains the health and safety of each satellite.

(iii) User Segment

The user segment consists of GPS receivers, processors and

antennas utilized for accurate positioning and timing by the users.

The user estimates the pseudo range (actual range + error) of each

satellite by measuring the transit time of the signal. The measured

transit time includes the actual travel time between the satellite and

receiver and the satellite clock bias between the satellite clock and

user clock. Using the pseudo ranges, the 3-D position (Latitude,

Longitude and height) of the user and the time offset between the

transmitter and receiver clocks can be estimated.

Let the user be at xu , yu and zu in earth fixed, earth centered

coordinate system and the satellite’s be at xi , yi and zi (where

i=1,2,3,4,..) in the same coordinate system as the user. Assume that

the user starts his clock at tu seconds, receives signal at ti (i=1,2,3,4,..)

seconds from satellite and Δt is the time offset between the user and

satellite. User’s position (in 3-D) and time offset are obtained by

solving the nonlinear equations.

��� − ���� + ��� − ���� + ��� − ���� = � ���− �� + ����, �= �, �, �, �, ..--- (1.1)

50

Where, ‘c’ is the free space velocity of electromagnetic signal in

m/s. A user needs a minimum of four satellites in view to estimate his

position (Three coordinates and time).

A geometrical view of the pseudorange ( ρi ) measurements and the

resulting equations to be solved for the user position and receiver

clock bias are shown in Fig.1.3 [28].

Fig.1.3. Geometry to determine the User position

The use of L band gives acceptable received signal powers with

reasonable satellite transmit power levels and earth coverage satellite

antenna patterns. The path loss is proportional to �� for an Omni

directional antenna and is less at L band frequencies. Therefore L

band is selected for GPS and dual frequencies permit ionospheric time

delay measurements [1].The signals L1 and L2 are coherently derived

from a 10.23 M Hz basic clock and are given by

(xu,yu,zu

) O

��

Re

SV1

(x1,y1,z1)

SV2

(x2,y2,z2

)

SV3

(x3,y3,z3

)

SVi

(xi,yi,zi)

��

�� ��

51

L1= 154 x 10.23 M Hz =1575.42 M Hz

L2= 120 x 10.23 M Hz =1227.60 M Hz

The power levels of L1 and L2 at the output of satellite transmitters

and at the input of receivers close to the earth are given in Table 1.1.

Table 1.1 Comparison of Power levels of L1 and L2 frequencies

Frequency C/A code (dBW) P-code (dBW)

Transmitted Received Transmitted Received

L1 26.8 -160 23.8 -163

L2 ---- ---- 19.7 -166

Pseudorange Measurements

The measure of range, or distance, between the satellite and

receiver is known as the pseudorange. For position determination of

an user, the ranges from the satellites to the receiver are required. To

measure the pseudorange, either precision Code (P – code) or coarse

acquisition Code (C/A – code) can be used [28]. Assume that, Satellite

clock and receiver clock, which control the signal generation, must be

synchronized for a moment. When the Pseudo Random Noise (PRN)

code is transmitted from the satellite, the receiver generates an exact

replica of the same code. The transmitted code by the satellite will be

received by the receiver after some time, which is equivalent to the

signal travel time in free space. By comparing the received code and

the code generated by the receiver, the signal travel time can be

computed by the receiver. The range between the satellite and receiver

52

can be obtained by multiplying the signal travel time with the velocity

of light. This procedure is depicted in Fig. 1.4 [28]. In practice, the

synchronization between the satellite and receiver clocks is not

possible exactly due to the error between the satellite and receiver

clocks [78].

Fig.1.4. Pseudo range measurements

1.7 Sources of GPS Signal Errors

The different sources which affect the ranging accuracy of GPS

signals are:

(i) Ephemeris Error

(ii) Satellite clock Error

(iii) Ionospheric Error

(iv) Tropospheric Error

(v) Multipath Error

(vi) Receiver Error

The brief description about each error source is explained in the

following lines.

∆�

Satellite code “String of 0’s and 1’s

Identical code generated in receiver

53

(i) Ephemeris Errors

The ephemeris error results whenever the GPS message signal does

not transmit the correct location of the satellite. This error can be

resolved into radial and tangential components. Out of these two,

tangential component is large. But it will not affect the ranging

accuracy of GPS to the same level. It is observed that for predictions

up to 24 hours, the rms ranging error attribute to ephemeris is 2.1 m

[1, 79].

(ii) Satellite Clock Errors

The GPS ranging accuracy depends on the predictability of satellite

clock. The satellite clock errors are same for both the C/A - code and

P – code users. GPS uses atomic clocks to minimize this clock error.

As per reports, the average clock error is 1-2 m [1].

(iii) Ionospheric Errors

The positional accuracy of GPS is predominantly affected by the

ionospheric variations [15, 80]. The ionosphere varies greatly with the

amount of radiation received from the sun. The normal variations are

due to the diurnal and seasonal effects and the abnormal variations

are mainly due to the Sudden Ionospheric Disturbances (SID),

Ionospheric Storms, sporadic E-layer reflections, Tides and Winds,

Sunspot cycle, Fadings, Whistlers etc. [20]. The SID is first observed

by Mongel and Dellinger, hence S.I.D. is often referred to as Mongel-

Dellinger effect [81]. The extent of solar disturbances is measured by a

method of sun-spot counting.

54

SID’s are caused due to sudden unpredictable appearance of solar

flares from the sun which are more likely during peak solar activity.

Areas of instability in the sun release high speed plasma with huge

amount of matter and energy, called as the coronal mass ejections

(CME’s) throughout the whole 11-year solar cycle. In due course

these solar CME’s reach the earth’s magnetosphere, causing

great disturbances in the earth’s magnetic field, and are called as

geomagnetic storms, observed by ground magnetic observatories

[25]. The CME’s take around 20 hours to reach the earth. This type

of disturbances may last from a few minutes to about an hour and

takes place simultaneously everywhere on the sunlit portions on the

globe. The intensity of disturbances tends to be peak in the region

where the sun’s radiation is perpendicular.

The free electrons in the ionosphere will affect the velocity of GPS

signals. In the absence of ionization, electromagnetic waves travel with

velocity of light from satellite to ground receiver. The presence of

ionization decreases the group velocity of propagation and hence

results in delay of GPS signals. The ionospheric time delay is

proportional to TEC and to inverse of the carrier frequency squared

(1/f2) [1].

The effects of TEC on radio waves are:

(a) Group path delay

(b) RF carrier phase advance

(c) Doppler shift

(d) Faraday rotation

55

(e) Refraction:

(f) Distortion of pulse waveforms

These effects are briefly explained in the following lines.

(a) Group path delay:

The dispersive ionosphere introduces a time delay in the GPS

signals. The ionospheric time delay ‘τ’ of the signals is proportional

to the TEC along the signal path and the frequency of the

propagated signals and is given by

�= ���.�× ����× �� � Seconds ----- (1.2)

where, c is the velocity of light in m/s, f is the frequency in Hz and

TEC in el/m2 [1].

(b) RF carrier phase advance:

The ionosphere changes the phase (Φ) of the carrier signal. In the

absence of ionosphere, the phase will be advanced and its increment

is given by [1]

�� = ��.�������

�. ���� Cycles ----- (1.3)

This effect is very important in determination of space object

velocities by means of range rate measurements.

(c) Doppler shift:

With frequency being the time derivative of phase, an additional

frequency shift results due to variation in TEC and is given by [1]

��� = ����

= ��.������ �

� �(���)

��� Hz ----- (1.4)

The time rate of change of TEC leads to Doppler shift errors.

56

(d) Faraday rotation:

The plane of polarization of the signal changes/rotates as it passes

through the ionosphere. The amount of rotation is linearly

proportional to the component of the magnetic field in the direction of

propagation. The Faraday rotation can be approximated as [1]

� = �.������

�� �� ��� � ��� = �1.885 �� ��� � Radians ----- (1.5)

To overcome the Faraday rotation effect, GPS signals are

transmitted with right-hand circular polarization [39].

(e) Refraction:

When a radio wave passes through the ionosphere, refraction or

bending of the wave occurs. The amount of refraction is proportional

to the density of ionization of the layer, the frequency of the radio

wave and the angle at which the wave enters the layer. The bending

produces an apparent elevation angle higher than the geometric

elevation.

The angular refraction may be expressed by [1]

�� = � ����������������������������� ��� ������

���

� Radians ----- (1.6)

where �� is the apparent elevation angle, R is the apparent range, ��

is computed from �� = (40.3/��) x TEC, r� is the earth’s radius and h�

is the height of the centroid of the TEC distribution, generally between

300 and 400 km. The geometry related to the above equation is

presented in Fig.1.5 [38].

57

o

Fig.1.5. Deviation of ray path due to angular refraction

(f) Distortion of pulse waveforms:

The dispersive nature of the ionosphere causes pulse distortion. It

produces a difference in pulse arrival time across the bandwidth Δf

and is given by [1]

Δt� = ��.�� ��� � ��

TEC = �.���� �

�� ��� s/Hz ----- (1.7)

The dispersion across the 20 MHz GPS bandwidth is normally small

and can be ignored.

In addition to the above effects, GPS receivers used in low-latitude

regions also suffer from rapid amplitude and phase fluctuations

known as scintillations. Scintillations occur when the satellite signal

ΔE

E0

re

S

APPARENT PATH

hi

DIRECT PATH

RAY PATH

HORIZON

CENTER OF EARTH

O = OBSERVATION SITE S = SATELLITE VEHICLE

58

travels through small-scale ionospheric irregularities, typically during

evening and night times in equatorial regions. Frequent scintillations

and high rates of change in TEC can cause loss of lock in receivers. At

these times, GPS users in low latitudes can experience decreased

levels of accuracy [58, 82-83]. All users will correct the raw pseudo

ranges for the ionospheric delay by different techniques. The ranging

error due to ionosphere is about 4 m [1].

(iv) Tropospheric Errors

The variations in temperature, pressure and humidity of

troposphere will contribute for variations in the velocity of GPS

signals. Correspondingly a ranging error of 0.7 m takes place [1].

(v) Multipath errors

The reflected signals which enter into the receiver may sometimes

mask the actual signal and this error is known as multipath error.

This error can be minimized with proper combination of antenna

cutoff angle and antenna location. ‘Narrow correlator’ receivers also

minimize the impact of multipath errors. The ranging error due to

multipath is about 1 m [1].

(vi) Receiver Errors

Modern receivers use reconstructed carrier to aid the code tracking

loops and give a precision of better than 0.3 m. Correspondingly, the

net ranging error due to receiver is less than 0.5 m.

Out of all the above, the ionosphere is the main source of range

and range rate errors for users of satellite based navigation systems. A

typical estimation of the error budget is presented in Table 1.2 [1].

59

Table 1.2 GPS Error Budget

Errors Error Budget

Ephemeris data 2.1 m

Satellite clock 2.1 m

Ionosphere 4.0 m

Troposphere 0.7 m

Multipath 1.4 m

Receiver Noise 0.5 m

TOTAL 10.8 m

1.8 GPS Aided GEO Augmented Navigation (GAGAN)

The required accuracy, availability and integrity, which cannot be

obtained with stand-alone GPS, can be obtained by augmenting the

GPS, which are very important for precision approach applications in

both civilian and defense sectors. The basic aim of GPS Aided GEO

Augmented Navigation (GAGAN) is to provide navigation for all phases

of flight over the Indian airspace and in the neighboring area

especially for strategic defense applications [14].

There are five important elements in GAGAN like any SBAS. They

are Reference Stations, Master Control Center, Land uplink Stations,

Geostationary Earth Orbit (GEO) pay load and user GNSS receivers.

The Basic system elements of GAGAN are shown in Fig. 1.6 [3, 84].

The implementation of GAGAN consists of three phases. They are

(i) Technical Demonstration System (TDS)

(ii) Initial Experimental Phase and

(iii) Final Operational Phase (FOP)

60

GAGAN USER

OFC LINKS

1,2

OFC LINKS

1,2

COMMUNICATION NETWORK

INMCC-I INMCC-II

INMCC-III INLUS-I C1, C5

UP LINK

INLUS-II C1, C5

UP LINK

INLUS-III C1, C5

UP LINK

INRES (1,..,N)

INRES (1,..,N)

GPS L1, L2

GPS L1, L2

GPS L1, L2

GPS L1, L2

GEO-1 L1, L5

GEO-1 L1, L5

GEO-1 L1, L5

61

The initial phase is the Technical Demonstration System (TDS)

using GPS constellation. In this phase, eight Indian Reference

Stations (INRES) are planned to be installed over the Indian region at

Ahmedabad, Bangalore,Delhi, Jammu, Kolkata, Thiruvananthapuram,

Portblair and Guawhati airports. One Indian Master Control center

(INMCC) and one Indian Navigation Land Up-Link Station (INLUS) are

planned at Bangalore [3]. Further, one GEO is planned to be launched

at location 820 East.

After first phase (TDS), redundancies will be provided to the space

segment, INMCC and INLUS to validate the system over the entire

Indian airspace. Additional augmentation will be worked out based on

these results. During the third phase (FOP), the system becomes

operational.

The present GAGAN configuration includes 15 Indian Reference

Stations (INRES), 2 Indian Master Control centers (INMCC) & 3 Indian

Land Uplink Stations (INLUS) and 2 GEO satellites (GSAT-8 & GSAT-

10). All INRES are integrated with redundant communication links

(which include 2 OFC & 2 VSAT links) to transfer data to INMCC.

GAGAN Final System Acceptance Test (FSAT) is successfully

completed on 16th-17th July 2012 [85-87].

1.9 Organization of the thesis

This thesis comprises of 6 chapters including introduction and

conclusions. The details of previous work related to the above

mentioned objectives, fundamentals of GPS, the error sources of GPS

signals and about the details of Indian navigation system “GAGAN”

62

are discussed in this chapter. The diurnal and seasonal variations of

TEC during the year 2003 are analyzed in chapter 2. The storm time

variations of TEC during major geomagnetic storms are presented in

chapter 3. The TEC values during the Halloween storm, 2003 are

compared with different prediction model results in chapter 4 to verify

their accuracy and suitability for low latitude stations. A neural

network based region specific TEC model has been developed using

back propagation algorithm in chapter 5. Conclusions are presented

in chapter 6 and important topics related to the thesis are mentioned

in the appendices.