4
Receiver and Antenna
Design
4.1 RECEIVER ARCHITECTURE
Although there are many variations in GPS receiver design, all receivers must
perform certain basic functions. We will now discuss these functions in detail, each
of which appears as a block in the diagram of the generic receiver shown in Fig. 4.1.
4.1.1 Radio-Frequency Stages (Front End)
The purpose of the receiver front end is to ®lter and amplify the incoming GPS
signal. As was pointed out earlier, the GPS signal power available at the receiver
antenna output terminals is extremely small and can easily be masked by inter-
ference from more powerful signals adjacent to the GPS passband. To make the
signal usable for digital processing at a later stage, RF ampli®cation in the receiver
front end provides as much as 35±55 dB of gain. Usually the front end will also
contain passband ®lters to reduce out-of-band interference without degradation of
the GPS signal waveform. The nominal bandwith of both the L
1
and L
2
GPS signals
is 20 MHz (Æ10 MHz on each side of the carrier), and sharp cutoff bandpass ®lters
are required for out-of-band signal rejection. However, the small ratio of passband
width to carrier frequency makes the design of such ®lters infeasible. Consequently,
®lters with wider skirts are commonly used as a ®rst stage of ®ltering, which also
helps to prevent front-end overloading by strong interference, and the sharp cutoff
®lters are used later after downconversion to intermediate frequencies (IFs).
80
Global Positioning Systems, Inertial Navigation, and Integration,
Mohinder S. Grewal, Lawrence R. Weill, Angus P. Andrews
Copyright # 2001John Wiley & Sons, Inc.
Print ISBN 0-471-35032-X Electronic ISBN 0-471-20071-9

synthesizer
Clocks Interrupts
Digitized IF signal
External inputs
(INS, altimeter,
Loran-C,
clock aiding)
Navigation outputs
(position, velocity, time,
fault, detection/isolation)
Navigation
processing
(includes Kalman
filtering)
Code aquisition/tracking
Carrier acquisition/tracking
Message bit synchronization
Navig tion message demodulation
Code/carrier pseudoranging
Delta-range measurements
A/D
converter
a
H/W & S/W Signal Proccessing
Fig. 4.1 Generic GPS receiver.
4.1 RECEIVER ARCHITECTURE
81
3. Conversion of the signal to a lower frequency makes the sampling of the signal
required for digital processing much more feasible.
Downconversion is accomplished by multiplying the GPS signal by a sinusoid

N  kT
e
B 4:1
where k  1:3806 Â 10
À23
J=K, B is the bandwidth in Hz, and T
e
is the effective
noise temperature in degrees Kelvin. The effective noise temperature is a function of
sky noise, antenna noise temperature, line losses, receiver noise temperature, and
ambient temperature. A typical effective noise temperature for a GPS receiver is
513 K, resulting in a noise power of about À138:5 dBW in a 2-MHz bandwidth and
À128:5 dBW in a 20-MHz bandwidth. The SNR is de®ned as the ratio of signal
power to noise power in the IF bandwidth, or the difference of these powers when
82
RECEIVER AND ANTENNA DESIGN
expressed in decibels. Using À154:6 dBW for the received signal power obtained in
Section 3.3, the SNR in a 20-MHz bandwidth is seen to be À154:6 À
À128:5À26:1dB. Although the GPS signal has a 20-MHz bandwidth, about
90% of the C=A-code power lies in a 2-MHz bandwith, so there is only about 0.5 dB
loss in signal power. Consequently the SNR in a 2-MHz bandwidth is
À154:6 À 0:5ÀÀ138:5À16:6 dB. In either case it is evident that the signal
is completely masked by noise. Further processing to elevate the signal above the
noise will be discussed subsequently.
4.1.3 Digitization
In modern GPS receivers digital signal processing is used to track the GPS signal,
make pseudorange and Doppler measurements, and demodulate the 50-bps data
stream. For this purpose the signal is sampled and digitized by an analog-to-digital
converter (ADC). In most receivers the ®nal IF signal is sampled, but in some the
®nal IF signal is converted down to an analog baseband signal prior to sampling. The

shift the signal to precisely zero frequency and phase. Because the shift to zero
frequency results in spectral foldover of the signal sidebands, both in-phase (I ) and a
quadrature (Q) baseband signal components are formed in order to prevent signal
information loss. The I component is generated by multiplying the digitized IF by
the NCO output and the Q component is formed by ®rst introducing a 90

phase lag
in the NCO output before multiplication. Feedback is accomplished by using the
measured baseband phase to control the NCO so that this phase is driven toward
zero. When this occurs, signal power is entirely in the I component, and the Q
component contains only noise. However, both components are necessary both in
order to measure the phase error for feedback and to provide full signal information
during acquisition when phase lock has not yet been achieved. The baseband phase
y
baseband
is de®ned by
y
baseband
 atan2I ; Q4:2
where atan2 is the four-quadrant arctangent function. The phase needed for feedback
is recovered from I and Q after despreading of the signal. When phase lock has been
achieved, the output of the NCO will match the incoming IF signal in both frequency
and phase but will generally have much less noise due to low-pass ®ltering used in
the feedback loop. Comparing the NCO phase to a reference derived from the
receiver reference oscillator provides the phase measurements needed for carrier
phase pseudoranging. Additionally, the cycles of the NCO output can be accumu-
lated to provide the raw data for Doppler, delta-range, and integrated Doppler
measurements.
Code Tracking and Signal Spectral Despreading The digitized IF signal,
which has a wide bandwidth due to the C=A- (or P-) code modulation, is completely

4.2.1.1 Receivers with Channel Time Sharing
Single-Channel Receivers In a single-channel receiver, all processing, such as
acquisition, data demodulation, and code and carrier tracking, is performed by a
single channel in which the signals from all observed satellites are time shared.
Although this reduces hardware complexity, the software required to manage the
time-sharing process can be quite complex, and there are also severe performance
penalties. The process of acquiring satellites can be very slow and requires a
juggling act to track already-acquired satellites while trying to acquire others. The
process is quite tricky when receiving ephemeris data from a satellite, since about
30 s of continuous reception is required. During this time the signals from other
satellites are eclipsed, and resumption of reliable tracking can be dif®cult.
After all satellites have been acquired and their ephemeris data stored, two basic
techniques can be used to track the satellite signals in a single-channel receiver. In
slow-sequencing designs the signal from each satellite is observed for a duration
(dwell time) on the order of 1s. Since a minimum of four satellites must typically be
observed, the signal from each satellite is eclipsed for an appreciable length of time.
For this reason, extra time must be allowed for signal reacquisition at the beginning
of each dwell interval. Continually having to reacquire the signal generally results in
less reliable operation, since the probability of losing a signal is considerably greater
as compared to the case of continuous tracking. This is especially critical in the
presence of dynamics, in which unpredictable user platform motion can take place
during signal eclipse. Generally the positioning and velocity accuracy is also
degraded in the presence of dynamics.
If a single-channel receiver does not have to accurately measure velocity, tracking
can be accomplished with only a frequency-lock loop (FLL) for carrier tracking.
Since a FLL typically has a wider pull-in range and a shorter pull-in time than a
phase-lock loop (PLL), reacquisition of the signal is relatively fast and the
sequencing dwell time can be as small as 0.25 s per satellite. Because loss of
phase lock is not an issue, this type of receiver is also more robust in the presence of
dynamics. On the other hand, if accurate velocity determination is required, a PLL

must therefore modify the calculations listed above to take this into account.
Because current technology makes the hardware costs of a multichannel receiver
almost as small as that for a single channel, the single-channel approach has been
almost entirely abandoned in modern designs.
Another method of time sharing that can be used in single-channel receivers is
multiplexing, in which the dwell time is much shorter, typically 5±10 ms per satellite.
Because the eclipse time is so short, the satellites do not need to be reacquired at
each dwell. However, a price is paid in that the effective SNR is signi®cantly reduced
in proportion to the number of satellites being tracked. Resistance to jamming is also
degraded by values of 7 dB or more. Additionally, the process of acquiring new
satellites without disruption is made more demanding because the acquisition search
must be broken into numerous short time intervals. Due to the rapidity with which
satellites are sequenced, a common practice with a two-channel receiver is to use a
86
RECEIVER AND ANTENNA DESIGN
full complement of PN code generators that run all the time, so that high-speed
multiplexing of a single code generator can be avoided.
Two-Channel Receivers The use of two channels permits the second channel to
be a ``roving'' channel, in which new satellites can be acquired and ephemeris data
collected while on the ®rst channel satellites can be tracked without slowdown in
position=velocity updates. However, the satellites must still be time shared on the
®rst channel. Thus the software must still perform the functions listed above and in
addition must be capable of inserting=deleting satellites from the sequencing cycle.
As with single-channel designs, either slow sequencing or multiplexing may be
used.
Receivers with Three to Five Channels In either slow-sequencing or multi-
plexed receivers, additional channels will generally permit better accuracy and
jamming immunity as well as more robust performance in the presence of dynamics.
A major breakthrough in receiver performance occurs with ®ve or more channels,
because four satellites can be simultaneously tracked without the need for time

certain applications.
Dual-Frequency Ionospheric Correction Because the pseudorange error
caused by the ionosphere is inversely proportional to the square of frequency, it
4.2 RECEIVER DESIGN CHOICES
87
can be calculated in military receivers by comparing the P-code pseudorange
measurements obtained on the L
1
and L
2
frequencies. After subtraction of the
calculated error from the pseudorange measurements, the residual error due to the
ionosphere is typically no more than a few meters as compared to an uncorrected
error of 5±30 m. Although civilians do not have access to the P-code, in differential
positioning applications the L
2
carrier phase can be extracted without decryption,
and the ionospheric error can then be estimated by comparing the L
1
and L
2
phase
measurements.
Improved Carrier Phase Ambiguity Resolution in High-Accuracy Differ-
ential Positioning High-precision receivers, such as those used in surveying,
use carrier phase measurements to obtain very precise pseudoranges. However, the
periodic nature of the carrier makes the measurements highly ambiguous. Therefore,
solution of the positioning equations yields a grid of possible positions separated by
distances on the order of one to four carrier wavelengths, depending on geometry.
Removal of the ambiguity is accomplished by using additional information in the

Because the C=A-code has only 1023 chips per period, it is relatively easy to
acquire. In military receivers direct acquisition of the P-code would be extremely
dif®cult and time consuming. For this reason these receivers ®rst acquire the C=A-
code on the L
1
frequency, allowing the 50-bps data stream to be recovered. The data
contains a hand-over word that tells the military receiver a range in which to search
for the P-code.
88
RECEIVER AND ANTENNA DESIGN