Background
Although the focus of this text is on the algorithms for software signal processing of the Global Navigation Satellite System (GNSS) signals, it is important to consider the source of that data stream to be processed. Since "software" signal processing is stated, it implies an input digital data stream. Thus, the purpose of this topic is to provide some insight into how the satellite signals propagating through space result in this digital data stream. This is done, of course, via a GNSS antenna/front end.The purpose of this topic is to illustrate functional designs for GNSS, discuss the tradeoffs associated with different designs, and provide a basic understanding of the source of the digital data to be processed. The focus is on the narrowband GNSS L1 signals, primarily the Global Positioning System (GPS) Coarse/Acquisition (C/A) code, but references are made to the Galileo BOC(1,1) code where appropriate. At the end of the topic, multiple-band GNSS front ends are introduced.
The process begins with the GNSS signal, propagating through space, which is incident on a user’s GNSS antenna. This, in turn, induces a voltage within the element. That voltage is extremely weak, corresponding to a guaranteed signal power of -160dBW in the case of the Global Positioning System (GPS) [see ICD-GPS-200 (1991)] and has a carrier frequency of 1575.42 MHz. Considering a bandwidth of 2 MHz (the approximate null-to-null bandwidth the GPS C/A code signal), the received GPS signal power is actually below that of the thermal noise floor, as defined by Equation (4.1) with a simplified illustration in Figure 4.1.
FIGURE 4.1. Frequency domain depiction of the GPS signal and thermal noise power. Remember that 30 dBm = 1 dBW. Center frequency 1575.42MHz.
Let the Boltzmann’s constant be denoted by
the absolute temperature by
and the equivalent noise bandwidth by B in Hz, then
For the GPS C/A code signal
can be approximated by
or more conveniently expressed in dB as
This is quite unique in the field of radio transmission. For example, if you connected a traditional GPS antenna to a spectrum analyzer and searched for the presence of the GPS signal, then any such characteristics of the signal would be hidden as the observation would be dominated by the thermal noise. This is a feature of the code division multiple access (CDMA) spread spectrum signal and requires the appropriate signal processing to acquire and process the signal. This also implies that the design of the front end is based more on the level of the thermal noise rather than the received L1 band navigation signal. Thus, the voltage induced within the GNSS antenna element results from the thermal noise, which dominates, as well as the GNSS signals from the satellites in view. Given that that L1 GNSS band is a designated Aeronautical Radio Navigation Service frequency band, no other signals should be present within the frequency span.
The analog voltage that results from the incident GPS signal and thermal noise remains much too weak and at too high a frequency for most analog-to-digital converters (ADCs) to operate. In order to overcome this, the front end will utilize a combination of amplifier(s), mixer(s), filter(s), and its own oscillator to condition the incident voltage on the antenna to the resulting digital samples.
A fully functional GNSS L1 front end is depicted in Figure 4.2. In the coming sections, the function of each of the elements within the figure will be discussed using this implementation as a case study.
FIGURE 4.2. GNSS L1 front end.
GNSS L1 Front-End Components
GNSS Antenna
The antenna is typically not considered part of the front-end design, but since it is the first component in the signal path and dictates elements that follow, it is important to summarize when describing the GNSS front end. There are numerous texts on antenna theory and design, e.g., Balanis (1996), Straw (2003). Also the trade publication GPS World, over the past four years, has published a GPS Antenna Survey that lists all GPS antennas and their features. All of these are excellent references for additional information on GNSS antennas.
As is the case with most of the components associated with analog signal conditioning, there is an extensive set of parameters associated with an antenna that describe its performance. Three fundamental parameters to be discussed here are the frequency/bandwidth, polarization, and gain pattern.
The antenna will be designed to induce a voltage from radio waves propagating at the GNSS L1 frequency or 1575.42MHz. In addition, the design should accommodate the appropriate bandwidth of the desired signal. This is usually specified using two additional antenna parameters: the Voltage Standing Wave Ratio (VSWR) and impedance. Practically all GNSS front-end components utilize an impedance of 50 which is typical for a majority of radio frequency design. VSWR is a measure of impedance mismatch or the measure of how much of the incident power will be absorbed and how much will be reflected. And, of course, this is a function of frequency. The VSWR is typically on the order of 2.0:1, which equates to 90% power absorption across the bandwidth of desired frequencies.
Polarization refers to the orientation of the electric field from the radio frequency transmission. Received GNSS signals are right-hand circularly polarized (RHCP), and the antenna should be designed as such. The decision to employ RHCP for GNSS was definitely not arbitrary. One of the most difficult error sources to mitigate for GNSS is multipath. When the GNSS signal is reflected off an object, an undesirable situation for a system attempting to measure time-of-flight, the polarization will flip to left-hand circular polarization (LHCP). An RHCP antenna is quite effective in suppressing the LHCP reflection and minimizing this error source. Of course, a second reflection will reestablish the RHCP polarization, but the signal power is also likely diminished as a result of the multiple reflections. Thus, the polarization of the GNSS antenna provides a significant level of suppression from erroneous multipath reflection.
The antenna pattern describes the directivity of the antenna. The most basic idea for the antenna pattern would be one that receives signals equally from all directions—this is known as an isotropic antenna. However, such a uniform gain pattern does not make sense for GNSS. Since the signal source, GNSS satellites, are overhead for most applications the preferred antenna pattern would be hemispherical, designed to receive signal from only positive elevation angles from all azimuth directions. Given the problem of multipath and that most multipath rays arrive from low elevation angles, the antenna pattern could be crafted such that it was designed to receive signals only above 10°-20° elevation. Such an approach is definitely bound to further reduce multipath reflections, but as a consequence, the low elevation satellite signals would also be neglected, decreasing the availability of satellite measurements. A promising research area within GNSS antennas is that of antenna arrays, or a combination of individual antenna elements combined in such a way to shape distinct antenna pattern beams and nulls. Such an implementation should provide significant performance enhancement for GNSS.
Probably two of the most popular GNSS L1 antenna implementations are the patch and helix approaches but others also exist. These refer to the actual construction of the antenna element itself. Yet the parameters above should provide a measure of comparable performance between antennas.
The last topic in regard to antennas refers to the choice of an active or passive antenna. An antenna will often be integrated with other front-end components that improve their performance or are necessitated by the environment in which the antennas will operate.
One important parameter in front-end design is the overall noise figure
of the system. This parameter quantifies the noise added as a result of the analog signal conditioning. Of course, any additive noise or decrease in signal-to-noise ratio (SNR) is undesirable and should be minimized.
Denoting the resulting system noise figure by
the noise figure
of the nth element in cascade, and the gain of the nth element in cascade
the formula for noise figure is
What this equation indicates is that the first element in the RF cascade dominates the resulting noise figure for the system. This indicates that all passive components (cables and filters) prior to the first amplifier will have a negative impact on the noise figure. Likewise, components that follow a high gain amplifier in the cascade will have a minimal effect on the overall noise figure.
For example, consider working with a GNSS receiver in a laboratory environment. The optimal position for the GNSS antenna will be on the rooftop, clear of any obstructions. In most cases, this will require a lengthy cable run to the GNSS receiver with its self-contained front end. This RF cable from the antenna to the front end will be the first component within the cascade of components. Since all RF cables have some degree of attenuation, or noise figure, and no gain, the system noise figure will be severely degraded. This can be improved if an amplifier can be incorporated within the antenna itself prior to the long cable. This implementation is the norm in many GNSS antennas, and such a design is known as an active antenna and is characterized by the gain of the amplifier.
This active antenna approach complicates things slightly as the antenna itself is now considered an active element and requires power for the internal amplifier. This is accomplished in most cases using a bias-tee. The bias-tee component has three ports: RF, RF+DC, and DC. This component injects DC power onto the antenna cable from the front end to power the amplifier within the antenna. Thus, the antenna cable is utilized to pass the GNSS signal from the antenna to the bulk of the analog signal conditioning and then also to provide a DC voltage from the analog signal conditioning to the amplifier within the antenna. This is the approach outlined in Figure 4.2.
A passive antenna is practical in those designs that have the antenna in close proximity to the analog signal conditioning and, in particular, the first amplifier. This is commonly the case in the handheld GNSS receivers or for configurations employing expensive low-loss RF cables.
Filter
The first component within the RF path is a filter. A filter is a frequency selective device that allows only certain frequencies to pass and attenuates others.
The treatment of the filter as well as the following individual components will be kept terse. It is expected the reader has a basic background in signal processing; this will allow the focus to be on the overall GNSS front-end design.
This first filter in Figure 4.2 is a bandpass filter, as opposed to a lowpass or highpass filter, and its purpose is to provide additional frequency selectivity. Ideally, the antenna would only induce voltages for precisely the frequency band of interest. However, the antenna, like practical filters, is not ideal. The ideal component would pass a range of frequencies and completely eliminate those frequency components outside that range. Unfortunately, such a filter does not exist, and the transition between those frequencies that are passed and removed is a gradual transition. Further, even signals at frequencies within the passband typically experience some level of attenuation.
Typical antennas have fairly poor frequency selectivity. When this is considered, along with the received signal power levels (and the amplification that will be required), it is important to try to eliminate any high-power, out-of-band signal sources that could enter the front end and saturate later-stage components.
FIGURE 4.3. Comparison of bandpass filters.
For this reason, a bandpass filter typically is the first component following the antenna. The other filters within Figure 4.2 serve specific roles, as will be discussed in the following sections, but all are used to perform the primary role of a filter: pass selective frequencies and attenuate others.
Filters can be characterized by their insertion loss, or the attenuation of the desired frequency components. Ideally, there would be no insertion loss, but alas that is not the case for practical components, and the lower this value the better. Note that this filter insertion loss will result in a system noise figure penalty when it is placed prior to the first amplifier. Yet this is still very often the case to minimize any issues from adjacent frequency bands given the limited selectivity of the antenna itself. If the receiver will be operated in an environment that will not have high-power adjacent frequency band transmitters in close proximity, then this filter may not be necessary and the position of the first amplifier and filter can be switched. Or the impact to the noise figure can be minimized by selecting a filter with a low insertion loss.
The second filter parameter is the bandwidth. Again, since no filter is ideal, typically a 3 dB bandwidth is specified. This indicates at which frequency(s) the attenuation will be 3 dB (or 50% of the signal power). However, these two parameters cannot completely describe most filters but only provide some insight into their performance, as shown in Figure 4.3.
A goal in filter design is to provide sharp transitions between the desired (pass-band) and undesired (cutoff) frequencies while maintaining a minimal insertion loss. Depending on the practical implementation of the filters [options include cavity, surface acoustic wave (SAW), ceramic, or lumped elements (resistors, capacitors, and filters)], this can be done by increasing the number of sections/elements within the design.
