Unlike active users of the radio spectrum, the passive services do not interfere with other passive users and therefore can share spectrum efficiently with each other. However, as the passive services are dependent on the absence of transmissions, they cannot make their presence known to active users. Thus, they require negotiations to reserve those portions of the radio-frequency (RF) environment that are critical to their needs and to resolve conflicts when that spectrum is infringed upon by active services. As discussed in the previous chapters, frequency allocations for the passive services generally include allocations for specific molecular and atomic transitions, and allocations spaced approximately every octave for continuum and Earth observations. However, under the fast-paced development of technology and the subsequent increasing appetite for wireless communications and radio distancing devices that were once under only the purview of specialized sectors of commercial, government and academic institutions, the radio environment has become increasingly crowded. An example of this is the 2010 Presidential memorandum that directed the Secretary of Commerce, working through the National Telecommunications and Information Administration, to collaborate with the Federal Communications Commission (FCC) to make available a total of 500 MHz of federal and nonfederal spectrum over the next 10 years for mobile and fixed wireless broadband use. More recently, in 2012, the President’s Council of Advisors on Science and Technology (PCAST) recommended that 1000 MHz of federal and non-federal spectrum be freed up over the next 10 years and shared with private industry in order to meet the country’s needs for mobile and fixed wireless broadband access. This increase in use of the radio spectrum will likely be geographically diverse and less regulated compared to previous practice when broadcast installations were more specialized and access to the spectrum more tightly controlled. Hence, the effects of RF interference (RFI) on the passive services will likely increase in the upcoming decade thus creating new challenges and hopefully offering new solutions for addressing this changing landscape.
Radio spectrum usage can act as a zero-sum game: use by one application can exclude use by another. The key issue in allocation and use is to avoid this conundrum and to find balance by utilizing the best tools at hand to maximize the benefits of spectrum use. The suite of tools can be very different in different parts of the spectrum, and will change as time goes on and as technology changes.
Early documentation of the negative consequences of radio interference to the passive use of the radio spectrum for radio astronomy was performed by the International Radio Consultative Committee (CCIR) of the International Telecommunication Union (ITU).1 The threshold levels of interference to radio astronomy bands are published in Recommendation ITU-R RA.769 (05/03), Protection Criteria for Radioastronomical Measurements,2 which provides recommendations and guidelines on protection of the spectrum for radio astronomical measurements. Protections such as these are increasingly important as airborne-, ground-, and space-based use of the RF spectrum increase and as the potential for aggregate interference increases due to the proliferation of electronic devices.
The threshold levels of interference detrimental to radio astronomy3 given in Recommendation ITU-R RA.769 are specified in both power flux density (pfd) and spectral power flux density (spfd) at the radio telescope site. They are based on a consideration of the effect of interference on measurements of the total power received in a single antenna. Several criteria are basic to this analysis:
- The maximum level of interference that can be tolerated is that which increases the output power of the receiver by 10 percent of the root mean square (rms) noise level at the output after averaging for 2000 seconds. Because of the low level of cosmic signals, radio astronomers almost always work at levels close to the limits set by the system noise. This criterion can be thought of as allowing uncertainties in the data to be increased by 10 percent.
- Interference that is being considered is received in the antenna sidelobes. It is not realistic to set thresholds for interference received in the main beam of a large radio telescope; such interference occurs only transiently and is dealt with by editing the data. A value for the sidelobe gain of 0 dBi (i.e., equal to that of an isotropic radiator) is used in the calculations.4 The corresponding collecting area is l2/4p,
1 The CCIR has been replaced by the ITU Radiocommunication Sector (ITU-R).
2 These thresholds were originally published in CCIR Report 224, Documents of the Xth Plenary Assembly, Volume IV, Geneva, 1963, p. 331.
3 For background information from technical papers regarding the detection of radio-frequency interference, see the following: E.G. Njoku, P. Ashcroft, T.K. Chan, and L. Li, Global survey and statistics of radio-frequency interference in AMSR-E land observations, IEEE Transactions on Geoscience and Remote Sensing 43(5):938-946, 2005; S.W. Ellingson and J.T. Johnson, A polarimetric survey of radio frequency interference in C- and X-bands in the continental United States using WindSat radiometry, IEEE Transactions on Geoscience and Remote Sensing 44(3):540-548, 2006; and J.T. Johnson, A.J. Gasiewski, B. Güner, G.A. Hampson, S.W. Ellingson, R. Krishnamachari, N. Niamsuwan, E. McIntyre, M. Klein, and V.Y. Leuski, Airborne radio frequency interference studies at C-band using a digital receiver, IEEE Transactions on Geoscience and Remote Sensing 44(7, Pt. 2):1974-1985.
4 The ability of an antenna/receiver system to detect a signal (or equivalently, the susceptibility of being adversely affected by signals) is dependent on the inherent sensitivity of the receiver, as well as on the direction that the antenna happens to be pointing relative to the transmitter. However, even though the antenna may not be pointing directly toward the transmitter, it nonetheless has some ability to receive signals from essentially all directions, which is roughly equivalent to the reception achieved by a small wire antenna. Such a wire antenna is approximately able to receive signals equally from all directions and is therefore called an isotropic antenna (although a true isotropic antenna is not realizable in practice). Because the reception is approximately the same, the off-axis performance is generally assumed to be equivalent on average. A ratio of 1 converts to 0 dBi, where the “i” refers to the fact that the comparison is being made to an isotropic antenna (see Appendix F for a full discussion of the 0 dBi sidelobe gain).
The use of 0 dBi means that the transmitter engineer need not consider the potential impact on a large variety of different radio telescope designs, the calculations are much simplified when gain and pointing direction are removed as variables, and the onus is on the radio astronomy design engineers and observers for dealing with regions of the radio astronomy antenna pattern near the main beam that are above isotropic. In some specific cases, for example, nongeosynchronous satellites, the 0 dBi model may not be adequate.
where l is the wavelength. The choice of the 0 dBi level is discussed in Chapter 4 of the ITU Handbook on Radio Astronomy (2013 edition),5 with respect to the models of sidelobe gain for large antennas that are given in Recommendations ITU-R SA.509, S.580, and S.1428.
- The signal-to-noise ratio at the output of the receiver is measured after the data values have been averaged for a period of 2000 seconds. This value was typical of a long integration period in the 1960s when these calculations were first made; it continues to be used as a generally representative value. More recently there are observations, such as searches for prebiotic interstellar molecules, that push sensitivity to the limits of what is possible, with integration times on the order of days and even months.
For an interfering signal with pfd FH, the interference-to-noise ratio can be calculated using the values of bandwidth, antenna noise temperature, and receiver noise temperature appropriate for each band.6 By equating the interference-to-noise ratio to 0.1, the threshold value FH is obtained. The corresponding value of the threshold as spfd across the band is SH = FH/Df, where Df is the bandwidth. In Tables 1 and 2 of Recommendation ITU-R RA.769, values of FH and SH are given for representative radio astronomy bands across the spectrum for both continuum and spectral line observations. Both criteria must be met, unless the band is designated exclusively for either continuum or spectral line use. For continuum observations, the bandwidth used is the width of the allocated radio astronomy band; for spectral line (multichannel) observations, a value for the channel bandwidth appropriate for observations within the particular band is used. These values apply to observations that measure the total power received in a single antenna. Threshold values for interferometers and synthesis arrays are somewhat less stringent and are considered in Section 4.3.5. However, such instruments are not suitable for observations of extended objects, for which total-power observations are essential. Thus, the values in Tables 1 and 2 of Recommendation ITU-R RA.769 are the basic protection criteria for radio astronomy.
In contrast to radio telescopes that point from Earth to space, Earth Exploration Satellite Service (EESS) sensors point in the opposite direction, from space to Earth. Hence, radio sources on Earth, particularly those transmitting toward space, and which may be benign to radio astronomers, are typically incompatible with spaceborne EESS sensors. The rapid motion of satellite-based sensors through space for EESS remote sensing limits the integration time available for sensor measurements to seconds, compared with the longer integration times used by stationary radio telescopes. Hence when interference does occur to the sensor, it is more difficult to correct or compensate and so the data is often flagged simply as being lost.
Because Earth-observing satellite sensors typically point their antennas toward Earth’s surface, they are also particularly sensitive to ground-based interference sources. While some of the effects of RFI can be improved through front-end filtering prior to amplification and higher gain antennas that are better able to geographically isolate interfering sources, fundamental limitations on satellite resources such as size, weight, and power make most of those solutions impractical to implement.
There are two basic levels of remote sensing that utilize the RF spectrum: (1) active remote sensing, which is typically associated with the use of radars that make use of the physics of microwave scattering to infer physical characteristics about the targets being imaged, and (2) passive remote sensing,
5 International Telecommunication Union, ITU Handbook on Radio Astronomy, 2nd ed., Geneva, Switzerland, 2013.
6 Appendices 3 and 4 of the ITU Handbook on Radio Astronomy provide a useful guide on converting units used by radio astronomers to those used by others in the radio communications sector. The IUCAF (the Scientific Committee on Frequency Allocations for Radio Astronomy and Space Science) has available many useful lectures on these topics on its website at http://www.iucaf.org/.
which uses thermal emissions, radiative transfer theory, and models for the scattering albedo and surface emissivity to infer dielectric and emission properties of the natural environment.
While both of these remote sensing applications are vulnerable to RFI, the passive remote sensing measurements are more vulnerable to interference because there is no established way to detect and reject data that are contaminated with low-level interference—that is, interference that cannot be differentiated from signals originating from background thermal emission. The propagation of undetected contaminated data into numerical weather- and climate-prediction models may have a significant destructive impact on the reliability and/or quality of weather forecasting in some cases. In other cases observations may be partially obscured or denied completely owing to strong out-of-band or weak in-band emissions affecting regional or broad-area measurements.
Criteria for threshold levels of interference for passive remote sensing are defined by limiting the maximum permissible interference power within a reference bandwidth. This interference level is determined by fixing the unwanted signal level below 20 percent of ∆P = k∆TeB, where ∆Te is the sensitivity of passive radiometric sensors, k is Boltzmann’s constant, and B is the receiver bandwidth. Recommendation ITU-R SA.1028 provides the performance criteria for satellite passive remote sensing, and ITU-R SA.1029 provides the interference criteria that are compatible with those performance objectives by defining the maximum permissible interference level within a reference bandwidth that is not necessarily the same as any particular sensor’s bandwidth.
The ITU further recommends that in shared frequency bands (except absorption bands), the availability of passive EESS sensor data should exceed 95 percent from all locations in the sensor service area in the case where the loss occurs randomly, and that it should exceed 99 percent from all locations in the case where the loss occurs systematically at the same locations. For three-dimensional measurements of atmospheric temperature of gas concentration or of water vapor measurements, the availability of data shall exceed 99.99 percent. In bands that are allocated to the EESS and other passive services on an exclusive basis, the Radio Regulations state that “all emissions shall be prohibited (RR 5.340).” See Section 3.1.1, “Passive Sensors,” in this handbook for more information.
For active microwave remote sensing, frequency and bandwidth requirements have also been studied in the ITU-R and can be found in Recommendation ITU-R SA.577.4. Performance criteria for active microwave remote sensors have been extensively studied and have been defined in terms of interfering power within a reference bandwidth. These recommendations can be found in ITU-R SA.1166.1.
When considering the regulation of signals that may spill into science service bands, account should be taken of how such signals will appear to the scientific instruments in question. Radio astronomy observations of transient phenomena generally are vulnerable, as are Earth-observing satellites that view geographic regions for only brief periods of time as the satellite passes overhead. In radio astronomy, ultrahigh-energy cosmic rays can be observed in the very high frequency (VHF) radio band, which can be interfered with by human-made sources such as ocean-wave radars. Giant pulses from pulsars share a similar time-frequency signature with chirp radar.7 Nontransient observations can be interfered with by intermodulation and harmonic products from VHF broadcast signals (those generated from both the transmitting and the receiving end). These interfering signals are routinely detected throughout the L-band and appear somewhat like spectral lines. They are not difficult to discriminate against, but their
7 See S.W. Ellingson and G.A. Hampson, Mitigation of radar interference in L-band radio astronomy, Astrophysical Journal Supplement Series 147(1):167-176, 2003.
removal is time-consuming and often frustrating. In other words, interference that can be recognized as such can be excised, resulting only in a reduction in efficiency. But most unwanted signals that are not easily recognizable can masquerade as valid scientific data.
Because it is not possible for radio astronomy to operate in frequency bands for which there are transmitting antennas within the line of sight, the sharing of primary radio astronomy bands with services using satellite downlink transmissions or aeronautical transmissions is avoided. The most serious cases of interference to radio astronomy during recent years have resulted from transmitters on satellites producing unwanted emissions that fall within radio astronomy bands.
An example of interference from a geostationary orbit (GEO) satellite in a band adjacent to a radio astronomy band is provided by a European television broadcast satellite transmitting in the Fixed Satellite Service band 10.7-10.95 GHz. A measured spectrum showed that at 10.7 GHz, the upper edge of a primary radio astronomy band, the spfd from the satellite was approximately 39 dB greater than the corresponding threshold value for continuum observations in Table 1 of Recommendation ITU-R RA.769. The resulting radiation into the 10.6-10.7 GHz radio band makes that band completely unusable for observations by the 100 m radio telescope at Effelsberg, Germany: for further details see Chapter 6 in the ITU Handbook on Radio Astronomy (2013 edition).
Interference from satellites in GEO presents a special problem, because a constellation of interfering satellites distributed along the orbit could preclude science observation within a band of sky centered on the orbit. The apparent declination of the orbit varies by approximately 10° as seen from observatories at intermediate latitudes in the Northern and Southern Hemispheres of Earth (see Figure 1 in Annex 1 of Recommendation ITU-R RA.517, or see ITU-R RA.611). Thus, the whole sky can be observed if observations can be made to within 5° of the orbit from observatories in both hemispheres. In the sidelobe model in Recommendation ITU-R SA.509, the sidelobe gain at 5° from the main-beam axis is 15 dBi, so values of the detrimental thresholds for such observations are 15 dB lower than those based on a sidelobe gain of 0 dBi, as in the tables in Recommendation ITU-R RA.769. It is desirable that these lower detrimental thresholds be applicable to unwanted emissions from GEO satellites. For further discussion, see ITU-R RA.769 or Chapter 4 of the ITU Handbook on Radio Astronomy (2013 edition).
Examples of interference from non-geostationary orbit satellites (non-GSO) can be found with the 24 satellite Russian Global Navigation Satellite System (GLONASS; 1602.5625-1615.5 MHz) and the 66 satellite Iridium constellation (1618.25-1626.5 MHz) that interfere with Radio Astronomy Service (RAS) operations in the 1610.6-1613.8 MHz and 1660-1670 MHz bands. Interferences of this type have been due to out of band emissions from lack of pulse shaping, poor control of modulation sidelobes in the frequency domain, and intermodulation products generated on the satellite caused by driving the transmitter amplifiers into compression in order to improve on efficiency. Because of their constant motion, constellations of non-GSO satellites such as these have the potential for being particularly disruptive to the passive services because of their constantly changing configuration and their near-constant coverage of Earth’s surface.
In the case of GLONASS, improvements have been made by no longer launching spacecraft with frequency capacity higher than 1610 MHz, and those that do launch, all are equipped with out-of-band filters. As such, after 1999, the out-of-band emissions from GLONASS in the 1612 MHz radio astronomy band have been eliminated. Unfortunately, this cannot be said of the Iridium system.
Recognizing that interference to radio astronomy is difficult to avoid at all times (see Figure 4.1), the ITU recommends that the fraction of time during which transmissions from any one service into a
FIGURE 4.1 Illustration of the effects of radio frequency interference (RFI) on a Radio Astronomy Service (RAS) VHF passive-only band taken by the Long Wavelength Array (LWA1; Taylor et al., 2012). Shown above are two pairs of images taken 10 seconds apart. Each row shows the radiometric intensity, I (left), and absolute circular polarization, |V| (right), for a part of the celestial sky that includes the Sun and Cas A, two very strong radio sources. As can be seen in the time stamp of the images in their upper left-hand corner, the top row of images was collected just 10 seconds prior to the bottom row. Effects of RFI in the protected RAS band from 73.0-74.6 MHz on the 100 kHz bandwidth images shown in the top row (centered at 74.0 MHz) completely obscure the strongest astronomical sources seen in the image, which are clearly visible in the bottom row. SOURCE: Courtesy of Greg Taylor and the Long Wavelength Array.
radio astronomy band exceeding the corresponding threshold level (ITU-R RA.769) should not exceed 2 percent. The corresponding fraction of time for the aggregate emission from all services within the radio astronomy band should not exceed 5 percent. These limits are specified in Recommendation ITU-R RA.1513, Annex 1 of which contains a discussion of the basis and application of the limits.
While these time restrictions are useful for fixed installations such as radio astronomical observatories, they are less useful for EESS radiometric observations, which have a strict linkage between the geographic-temporal aspects of coverage due to the fundamental mechanics of satellite orbits. Such sources of RFI can be particularly problematic because the antennas of the Earth-observing satellites are pointed such that maximum gain is obtained from Earth’s surface, and hence, ground-based sources of interference will have a maximal impact on the satellite instrument during the time that it is observing the geographic region where the interferer is located (Figure 4.2)
Several techniques enable active and passive users of the radio spectrum to share this limited resource. First, multiple users may share the same spectral allocation provided that they are spatially located at a sufficient distance, or with sufficient blockage (e.g., mountains), such that the line-of-sight propagation falls below threshold limits. Second, spectral allocations can recognize the incompatible nature of the various services and designate some spectral regions as protected for passive service use only. Finally, multiple users may effectively share the same spectral allocation if none need to use it continuously and all are able to coordinate the use of the spectral allocation based on local demand. Thus, while sharing of the radio spectrum is often considered a zero sum game, where use by one service excludes use by another, effective spectrum management includes consideration of location and time, as well as the specific spectral allocation.
The signal levels received by radio astronomers from cosmic objects are generally many tens of decibels weaker than the signal levels usually required for communications, broadcasting, radar, and other transmitting services. Earth stations in both EESS and Space Research Service receive weak signals: EESS from Earth-orbiting satellites and the deep space network from interplanetary spacecraft with extremely low signal levels. An important degree of protection from ground-based transmitters can be obtained by choosing observatory sites and Earth stations in locations of low population density and taking advantage of shielding by mountains or other terrain features. However, at frequencies above about 60 GHz, atmospheric absorption becomes important, and observatories must be located at high elevations. In these cases sites with effective terrain shielding are difficult to find. Transmitters on aircraft, spacecraft, balloons, and high-altitude platform stations can remain within the line of sight over long distances. Choice of the site for an observatory is of little help in providing protection against them.
It is useful to establish a coordination zone around an observatory for protection against terrestrial transmissions of a particular service. For example, coordination zones are used for cases in which the radio astronomy band is shared with the terrestrial mobile service (see Box 4.1). Such a zone can be defined by the requirement that, at the observatory, the sum total of all transmissions from outside the zone should not exceed a detrimental threshold for the frequency band concerned. Coordination zones can also be effective for protection in cases in which the potentially interfering transmissions are from a service with an allocation close to a radio astronomy band edge.
FIGURE 4.2 Percentage of radio frequency interference (RFI)-flagged samples in the passive and active bands of the radio spectrum. The top image shows the passive part of the spectrum between 1400-1427 MHz, which is a protected passive band. The lower image is collected at 1.26 GHz and shows the effects of ground-based radar stations on the observations. Ground-based RFI such as that shown here cause a degradation or loss of data that are important for weather, food security, and climate-prediction models that are supported by the Earth Exploration Satellite Service. SOURCE: Data shown above are from NASA’s Aquarius satellite, launched in 2011. Image courtesy of David LeVine, NASA Goddard Space Flight Center.
Defining the boundaries of a zone requires knowledge of the radiated powers in the direction of the observatory or Earth station and a determination of the transmission loss in various directions with respect to the observatory or Earth station. The transmission loss for propagation over terrain is calculable from terrain profiles. Major observatories and Earth stations usually have gridded elevation data for their surrounding area and software for the calculation of propagation loss. In the case of terrestrial mobile transmitters, it is often useful to use the Monte Carlo method to determine how the flux density at the observatory or Earth station varies with the size of the coordination zone, using the criteria in Recommendation ITU-R RA.1513 for the percentages of time for which flux density in excess of the detrimental thresholds can be allowed. If it is necessary that a transmitter be sited within the coordination
US385 Radio astronomy observations may be made in the bands 1350-1400 MHz, 1718.8-1722.2 MHz, and 4950-4990 MHz on an unprotected basis, and in the band 2655-2690 MHz on a secondary basis, at the following radio astronomy observatories:
Allen Telescope Array, Hat Creek, California: Rectangle between latitudes 40° 00′ N and 42° 00′ N and between longitudes 120° 15′ W and 122° 15′ W
NASA Goldstone Deep Space Communications Complex, Goldstone, California: 80 kilometers (50 mile) radius centered on latitude 35° 18′ N, longitude 116° 54′ W
National Astronomy and Ionosphere Center, Arecibo, Puerto Rico: Rectangle between latitudes 17° 30′ N and 19° 00′ N and between longitudes 65° 10′ W and 68° 00′ W
National Radio Astronomy Observatory, Socorro, New Mexico: Rectangle between latitudes 32° 30′ N and 35° 30′ N and between longitudes 106° 00′ W and 109° 00′ W
National Radio Astronomy Observatory, Green Bank, West Virginia: Rectangle between latitudes 37° 30′ N and 39° 15′ N and between longitudes 78° 30′ W and 80° 30′ W
National Radio Astronomy Observatory Very Long Baseline Array Stations: 80 kilometers (50 miles) radius centered on:
|Latitude (North)||Longitude (West)|
|Brewster, WA||48° 08′||119° 41′|
|Fort Davis, TX||30° 38′||103° 57′|
|Hancock, NH||42° 56′||71° 59′|
|Kitt Peak, AZ||31° 57′||111° 37′|
|Los Alamos, NM||35° 47′||106° 15′|
|Mauna Kea, HI||19° 48′||155° 27′|
|North Liberty, IA||41° 46′||91° 34′|
|Owens Valley, CA||37° 14′||118° 17′|
|Pie Town, NM||34° 18′||108° 07′|
|Saint Croix, VI||17° 45′||64° 35′|
Owens Valley Radio Observatory, Big Pine, California: Two contiguous rectangles, one latitudes 36° 00′ N and 37° 00′ N and between longitudes 117° 40′ W and 118° 30′ W and the second between latitudes 37° 00′ N and 38° 00′ N and between longitudes 118° 00′ W and 118° 50′ W
“In the bands 1350-1400 MHz and 4950-4990 MHz, every practicable effort will be made to avoid the assignment of frequencies to stations in the fixed and mobile services that could interfere with radio astronomy observations within the geographic areas given above. In addition, every practicable effort will be made to avoid assignment of frequencies in these bands to stations in the aeronautical mobile service which operate outside of those geographic areas, but which may cause harmful interference to the listed observatories. Should such assignments result in harmful interference to these observatories, the situation will be remedied to the extent practicable.”
SOURCE: Reprinted from FCC Online Table of Frequency Allocations, 47 C.F.R. 2.106, May 2015.
zone, then coordination between the transmitting service and the observatory/Earth station is required. Protection of the observatory or Earth station may be possible by various means, such as placing a null in the pattern of the transmitter in the direction of the observatory or Earth station or, if the allocated transmitter frequency is outside the radio astronomy band or spacecraft communication band, by additional filtering at the transmitter.
The calculation of separation distances for services in shared bands and the use of coordination zones are discussed in Annex 1 of Recommendation ITU-R RA.1031 and, for observations above 60 GHz, in ITU-R RA.1272.
A second strategy for mitigating the detrimental effect of an emission on a passive service is that of spectral separation. In this strategy, the center frequencies of the active-service emissions are moved sufficiently far from the band of interest of the passive service so as to reduce detrimental effects to an acceptable level.
The required separation depends on the strength and spectrum of the emission as seen by the receiver, as well as on the design of the receiver. All emissions naturally have spectra that extend above and below the carrier frequency. Regardless of how this emission is filtered before transmission, some fraction of this power remains. The total power that is subsequently delivered to a band used by passive services then depends on the transmitting power and the characteristics of the transmitter’s filter.
The design of the receiver determines the vulnerability of the receiver to out-of-band emissions through a separate mechanism: compression. Just as in the design of communications receivers, receivers for passive services represent a trade-off between sensitivity and linearity. The extremely weak nature of the signals of interest to scientific users constrains sensitivity requirements, with the result that a low-noise figure is essential. This consideration subsequently limits the type of filtering that can be done at the receiver. For example, any filter preceding the first amplifier must have sufficiently low insertion loss so as not to dominate the noise figure.
This constraint on the filter’s insertion loss limits the order of the filter design, which in turn limits the rate of attenuation with increasing frequency separation. Thus, the need for high sensitivity in the passive scientific services limits the amount of suppression that can be achieved for out-of-band signals. Just as in a communications receiver, out-of-band signals stronger than a threshold determined by this suppression cause the receiver to enter a nonlinear mode of operation—that is, compression. At this point, received signals become distorted.
In summary, there are two technical considerations in determining the separation between active and passive services required to limit interference to acceptable levels: (1) the actual out-of-band power spectral density delivered to the receiver and (2) the receiver’s limited ability to maintain linearity in the presence of strong out-of-band emissions, owing to the need to achieve an extremely low noise figure. There is no substitute for a situation-specific engineering analysis to determine the actual potential for detrimental interference due to these mechanisms; nonetheless, the situation is clearly helped by preferentially avoiding the assignment of bands used by high-power transmitters near the band edges of passive services.
A third strategy for mitigating interference between active and passive users of the radio spectrum is with dynamic frequency allocations based on local demand. For example, with increasing demand for
access to the radio spectra by commercial services (e.g., the communications industry) and a freeing up of the part of the spectrum once occupied by analog broadcast television (from 470-790 MHz), there has been a significant degree of activity in finding innovative ways for better accessing the RF spectrum in order to allow autonomous users to make use of the spectrum without interfering with one another and in compliance with FCC regulations. One example of this type of activity is in the area of cognitive radios that are meant to sense the local radio environment and dynamically create a communication channel that does not interfere with other users in that part of the spectrum.
In order to make best use of the spectrum, cognitive radios and other users wishing to access radio frequencies require knowledge of all current users in the geographic area. The Internet provides a unique opportunity to serve as the information backbone that supports this increased efficiency. For example, the IEEE Geosciences and Remote Sensing Technical Committee on Frequency Allocations in Remote Sensing (FARS) has created a geographic database that can be accessed through the Internet that combines satellite observations of the RF spectrum with ITU and FCC regulations. Other efforts being encouraged by the FCC result in commercially run databases that dynamically map the TV white space spectrum and can be used by cognitive radios for improving communications performance. Taken to its logical extension, databases such as these will likely be combined in the future and cover the wider RF spectrum to provide an up-to-date picture of spectrum use and provide a mechanism for dynamically allocating licenses to use the spectrum in accordance with ITU and FCC regulations and on a non-interfering basis. Such a database will be useful for scientific, governmental, and commercial users of the spectrum because it will allow for a more flexible approach to accessing parts of the spectrum that have been nominally precluded from use because of more traditional methods for spectrum licensing.
In considering interference from transmitters in other bands, definitions of the following terms, which can be found in Appendix A, are important: necessary bandwidth, out-of-band emission (OOBE), spurious emission, unwanted emissions.8 In addition, OOBEs are usually defined as those that fall within ±250 percent of the necessary bandwidth from the band center. With a few exceptions, spurious emissions are those at frequencies that are separated from the band center by more than ±250 percent of the necessary bandwidth. Further discussion of these definitions can be found in Chapter 6 of the ITU Handbook on Radio Astronomy (2013 edition).
Interference from transmitters in other bands can be generated by several mechanisms:
- Modulation sidebands can fall within a neighboring radio astronomy band when the transmitted spectrum is not adequately filtered.
- Two or more strong signals in a system that has a nonlinear response can generate beat frequencies at sums and differences of the frequencies and their harmonics.
- Harmonics can be generated by nonlinear responses within the transmitter, and filtering of the output to remove such unwanted responses may be absent or inadequate.
Limits on spurious emissions are given in Appendix 3 of the ITU Radio Regulations, and also in Recommendation ITU-R SM.329 (spectrum-management series). However, these limits generally fall
8 See F.H. Briggs and J. Kocz, Overview of technical approaches to radio frequency interference mitigation, Radio Science 40(5), 2005, and other articles in this issue; S.W. Ellingson, RFI mitigation and the SKA, Experimental Astronomy 17(1-3):261-267, 2004; R. Weber, G. Hellbourg, C. Dumez-Viou, A.J. Boonstra, S. Torchinsky, C. Capdessus, and K. Abed-Meriam, RFI mitigation in radio astronomy: An overview, Les Journees Scientifiques d’URSI-France L’electromagnetisme, Paris, France, 2013.
short of being able to protect radio astronomy to the required levels (Recommendation ITU-R RA.769) by several tens of decibels. Further discussion is given in Chapter 6 of the ITU Handbook on Radio Astronomy (2013 edition), including calculations based on these limits as applied to non-GSO and GEO satellites.
A basic modulation scheme used in communications is digital phase-shift keying. In its simplest implementation, the power spectrum is in the form of [sin(πDf)/πDf]2, where Df is a measure of the frequency relative to the center frequency of the transmitter. This type of spectrum has extensive sidelobes, which fall off with frequency by only 6 dB per octave of Df. For many years these sidelobes in GLONASS (Section 4.1.3) caused serious interference to observations in the 1610.6-1613.8 MHz and 1660-1670 MHz radio astronomy bands. One method of control of the sidebands of these and similar types of digital modulation is through the use of modulation techniques that control the rate of change of amplitude or phase at the transitions and thereby greatly reduce the level of unwanted frequency components that cause the sidebands. Several practical modulation systems of this type have been developed—for example, Gaussian-filtered minimum shift keying. These are capable of reducing the power level in the sidebands by many tens of decibels.
As explained in §4.2.2, the ability of passive services to employ filtering of sufficiently high order to mitigate the deleterious effects of emissions outside the radio astronomy band is severely limited by the need to achieve an extremely low noise figure. The requirements for a filter designed to mitigate out-of-band emission effectively must take into account (1) the strength of the interfering signal at the receiver, (2) the total received power level at which the receiver enters a state of compression, and (3) the potential for the generation of harmonic and intermodulation products due to any power remaining after filtering. It is not possible or reasonable to assign a single requirement for out-of-band suppression achieved by a filter for this application; situation-specific engineering analysis is required.
It should be noted that in certain cases it is possible to do better by employing alternative filtering technologies. For example, it may be more effective to notch specific interferers than to develop filters with broad band-stop characteristics. Also, it is now possible to design filters using cooled or superconducting materials that allow greater out-of-band suppression to be achieved without compromising insertion loss (sensitivity), albeit at greatly increased cost. As stated above, a proper situation-specific engineering analysis is required to determine whether these alternative techniques can provide the necessary attenuation.
Band-pass filtering in transmitters to confine the transmitted power to the allocated band of the transmitting service can be an effective technique in reducing band-edge interference. In certain cases it introduces technical difficulties. For example, in very high power transmitters, as used in some terrestrial television stations, the power dissipated in an output filter is a serious consideration. In some satellite transmitters, such as those of the Iridium system, the transmitting antenna is a phased array in which each radiating element is driven by a separate power amplifier. Power limitations on satellites necessitate the use of power amplifiers in which linearity is sacrificed to some degree to obtain high efficiency. Because harmonics and intermodulation products can be generated by nonlinearity in the output stages,
effective filtering requires a separate filter for each amplifier. The additional weight of such filters can make their implementation impractical. Filtering is more practical in satellites that use, for example, a parabolic reflector antenna in which the filtering is required in only one (or a small number) of signals going to the feed system.
In the case of transmitters in fixed terrestrial locations, it is often possible to place a null in the radiation pattern in the direction of a radio astronomy observatory. This technique has been used on numerous occasions to protect radio telescopes at Green Bank, West Virginia, within the National Radio Quiet Zone. In the case of a simple antenna such as a vertical dipole, generating the null basically requires using two such dipoles, fed through a power divider, and phased so that the two signals cancel for propagation in the direction of the observatory.
For a radio telescope that consists of a combination of individual antennas, the response to interference is generally reduced relative to that of a system that measures the total power received in a single antenna. The simplest combination is a two-element interferometer: that is, a system with two spaced antennas connected to a receiving system that produces an output proportional to the voltage product of the two received signals. Synthesis arrays consist of arrays of antennas connected in pairs as interferometers.
Two effects reduce the response to interference in interferometric systems. These are related to the fringe oscillations that occur as the relative phase of the signal from the two antennas varies, and, for geographically distributed systems, to the decorrelation that results from the difference in the time delays of the interfering signals in reaching the two antennas.
The treatment of interference in these cases is more complicated than for a single antenna. Results for two synthesis arrays and for very long baseline interferometry (VLBI) are plotted as a function of frequency in Recommendation ITU-R RA.769 and in the ITU Handbook on Radio Astronomy (2013 edition, Figure 4.2). These results are expressed in terms of the detrimental threshold levels for interference. The period of the fringe oscillations increases as the baseline (spacing) of the antennas, measured in wavelengths, is increased. The reduction in the interference response becomes effective when the fringe period is similar to or less than the time for which each data point is averaged. The reduction increases with frequency, because the fringe periods decrease. As the interference response is decreased, the corresponding detrimental threshold is increased.
For example, for the VLA in its longest-spacing configuration, the detrimental threshold at meter wavelengths is approximately 12 dB higher than that for a single antenna, and at 43 GHz it is 35 dB higher. For the VLBI, in which the antenna spacings are typically 100 km or more, the reduction of the fringe responses is effectively complete, but other effects of the interference result in detrimental thresholds no more than 40 to 50 dB higher than the corresponding single-antenna values.
It is important to note that observations using interferometric techniques are applicable to measurements of discrete sources, and the VLBI is only applicable to exceedingly small, bright objects. Total-power measurements using single antennas remain essential for measurements of extended sources and do not enjoy this advantage. Also, the ability of interferometric arrays to discriminate against interference decreases as the antennas become more closely spaced (as required for measurements of extended sources) and as the frequency goes down.
Increased occurrences of RFI have been noted by passive remote sensing in regions of the spectrum that are heavily used by other services. For example, airborne radiometers operating within the EESS-allocated 1400-1427 MHz band have incurred interference from radiolocation services operating adjacent to the band. Like RAS, passive remote sensing is also effected by out-of-band emission (as discussed in Section 4.3). Most satellite-based remote sensing applications are not stationary however and hence do not observe a fixed geographic region for extended periods of time. For this reason there is more tolerance for interferers in EESS than RAS, but the geographic reach of these sensors is global (or nearly so) by design and therefore the effect of OOBE and other sources of interference must be controlled over much larger geographic areas.
In the C-band region (5-7 GHz), NASA’s Earth Observing System Advanced Microwave Scanning Radiometer on Aqua is the first spaceborne radiometer with channels near 6.8 GHz since the Nimbus 7 Scanning Multi-channel Microwave Radiometer that stopped operating in September 1987. There is no EESS allocation near 6.8 GHz, and subsequent use of this spectral region, primarily by the fixed service, mobiles service, and fixed-satellite service, in accordance with the national and international allocations, has caused the band to be unusable for passive microwave measurements over land in much of the world’s developed regions without the application of RFI mitigation. Over the oceans, the ITU recognizes the passive microwave EESS usage between 6.425 and 7.075 GHz through Radio Regulation 5.458, but offers no formal protection other than asking administrations to keep the EESS needs in mind.
RFI mitigation techniques are being explored for passive radiometric observations in RFI-affected areas. In general, these techniques require costly and extensive modifications to the sensor and retrieval algorithms.
The technique of sub-banding requires the sensor’s receiver passband to be split into multiple sub-bands. Calibrated brightness temperatures are derived for each sub-band and are intercompared to identify any sub-bands that may have brightness temperatures exceeding a predetermined variance from the group. A logic tree is followed based on the level of agreement in brightness temperatures among the sub-bands. In cases where one sub-band is determined to be an outlier, it can be thrown out and the others averaged to provide an estimate of brightness temperature with only degradation in noise performance (sensitivity).
The digitization and signal-excision technique requires the signal passband to be digitally sampled. Initially this procedure was developed to be effective against RFI from pulsed emitters. After digital sampling, the time sequence of samples is checked for large deviations that are characteristic of coherent RFI. If the pulse repetition rate and pulse duration of the interferer are known, then the input can be effectively blanked during the pulse period with little impact to the radiometric sensitivity and accuracy. It should be noted however that narrow-band interfering signals will be broad in time, and thus are difficult to remove using this technique.
In general, a 1 to 2 percent bandwidth (which is the norm for most RAS allocations) is the minimum practical allocation for passive use, as dictated by the ability to filter OOBE from the receiver front-ends. However, given the impracticality of implementing narrowband filters that provide high transmission and significant out-of-band suppression, such as superconducting filters, a 5 percent bandwidth is preferred for the continuum bands. This need is strongly reinforced by the new and rapidly increasing requirements for bandwidth allocation at all frequencies, and it addresses interference concerns for both EESS and radio astronomy applications. For example, this requirement can be met by the use of the same fractional bandwidth allocations for spectral lines (such as for complex molecular line studies in the Milky Way and for redshifted lines of distant galaxies) as well as for continuum astronomy, so long as the allocated bands occur reasonably frequently throughout the full spectrum.
In addition, access to bandwidths much greater than the FCC-allocated bands are needed to exploit the capabilities of modern radio telescopes for both continuum and spectral line observations. At UHF and higher frequencies, propagation is limited to only a few multiples of the line of sight distance. Because radio telescopes are generally located in remote areas and are somewhat shielded by the surrounding terrain from ground-based RFI, protection to the RAS while needed over very broad bands, is only necessary for relatively small geographic areas surrounding each operational radio telescope. While radio astronomy needs protection over large segments of the radio spectrum, some degree of time sharing with active users, enabled by the Internet and dynamic frequency allocation, may be possible to make more efficient use of the overall spectrum (See Section 4.2). The problem is similar for EESS, although the requirements for sharing are different. In particular, Earth-viewing sensors in low Earth orbits map the entire globe, but may need only a few seconds of spectrum every other day for a given location. EESS sensors in geostationary orbit view the same area on the surface continuously, and time-sharing may not be feasible. Thus, while reserved frequency allocations are still a critical component for scientific use of the radio spectrum, where shared use of the spectrum is necessary, additional approaches must be considered.
The EESS and RAS communities share mutual concerns regarding unwanted emissions. In particular, passive EESS observations are prone to the same problems that face the radio astronomy community, although the two communities are usually looking in opposite directions (one looking to, the other from Earth). Strong efforts must be made to protect radio astronomy bands from interference due to air- or space-to-ground transmissions in other bands.9 Passive services are particularly sensitive to spurious, out-of-band, and harmonic emissions from other services. “Guard bands” proportional to frequency (or fractional bandwidth specified) should be maintained around the passive and shared bands, prohibiting services with an inherent inability to adequately filter transmissions from leaking into adjacent bands. A major effort to modernize and upgrade engineering standards for active services is desirable, especially with regard to out-of-band emissions. Modernization of these standards would be useful to other services as well as to RAS and EESS. This is particularly the case with airborne and satellite transmitters because of the potential clear line of sight between a passive user and unlicensed devices. As technology
9 See Appendix A for ITU definitions relating to interference.
improves, new applications in RAS and EESS will require lower noise floors and will benefit from new and wider bandwidth, which could be made available by improved rules for sharing.
In the past 30 years, radio astronomy studies have demonstrated the presence of ever-more-complex molecules in interstellar space. These discoveries have been one of the most fascinating and puzzling developments in the field. The molecular weight of the largest molecules already exceeds that of simple amino acids. It is anticipated that in the future, still-more-complex molecules, and possibly amino acids, will be found. The identification of complex molecules can be made only by the detection of a number of radio lines.
EESS environmental products (e.g., temperature and atmospheric pressure profiles) derived from passive radiometry would benefit from additional frequencies in the millimeter-wave bands. Specifically, additional allocations in the millimeter-band will permit multi-band passive radiometry to probe different depths of the atmosphere because of the varying penetration depths. Consequently, observations are desired either in, or adjacent to, bands allocated to other services. If unwanted emissions are minimized, the new science will be enabled.
At the other extreme of the frequency spectrum, the European Space Agency (ESA) is planning to launch the BIOMASS mission (432-438 MHz) in 2020 to map the biomass and carbon stored in the world’s forests. The longer wavelength at this frequency allows for better penetration into the vegetation canopy and hence a means for estimating the total forest volume. Because Europe and North America operate a missile warning and space-surveillance system in this band, this $500 million mission will not collect data in this part of the world. If some coordination between the satellite and these systems were possible, even on a very limited basis during the brief overpass of the satellite over these installations, then new observations over these geographic regions will be possible.
Although the advance of technology and widespread use of the spectrum has created challenges for passive users, it offers the potential to solve long-term RFI issues. The telecommunications industry and the FCC have been experimenting with dynamic allocation of bandwidth. In 2014, the FCC held a workshop on the Spectrum Access System, a pilot program where the spectrum at 3.5 GHz is dynamically scheduled based on priorities such as demand from primary and secondary users, geographic location, and transmitted power. A database of licensed use is maintained, and, at any given time, an unlicensed user may query the database and, if the band is unoccupied, may transmit in the band. The importance of this development cannot be understated, because it replaces the licensed use process, which has inherent inefficiencies—particularly regarding disputes between users—with one where permitted use is managed by software. Although this system is still in its infant stage and many important details need to be worked out, the potential to expand the use well beyond the 3.5 GHz band is intriguing. An observer wishing to utilize a previously unprotected band, could receive the required protection for the time needed to observe. Examples of this might include an observatory that wants to use a band dedicated to a space-Earth telecom service, or an EESS satellite that needs a clear band only when overhead. It is clear that this approach has the potential to revolutionize spectral access by taking advantage of the most important aspect of the spectrum as a resource—its nearly instantaneous renewability.
However, frequency coordination must take into account the rapid timescale at which radio telescopes operate. Most modern radio telescopes are equipped with multiple receivers and instruments,
spanning a wide range of radio frequencies. In the cases of telescopes operating at higher frequencies where clear, dry skies under stable nighttime conditions may be required for highest quality data, observatories have been employing dynamic scheduling. This enables the use of the telescope under the best possible conditions for the frequency being observed. Because weather conditions may change rapidly, a telescope schedule may change rapidly to observing projects requiring stronger degrees of spectrum protection. Even when weather conditions are stable, instrument failures or newly discovered hazardous asteroids may motivate changes in set observing schedules.
An additional factor in time-sensitive spectrum use is the coordination of simultaneous observations between two or more facilities. Radar observations of the moon or near-Earth asteroids transmit a radar signal from one station and receive at another, a bistatic operation. Because radar signal returns decrease rapidly in strength with more distant objects, the receiving station is highly vulnerable to RFI. Coordination with spacecraft, for radio science investigations, space-to-ground bistatic experiments, or for detections of safe landings or homing signals, also requires very specific timing of observations. Similarly, VLBI observations use a variety of telescopes worldwide and require interference-free access to the radio spectrum at all locations simultaneously. Thus, dynamic scheduling and frequency coordination must take into account both the local spectral environment and also those of the associated facilities in order to enable productive scientific use of the radio spectrum.