4
Technology and Opportunities for the Mitigation of Radio Frequency Interference

The capacity to address interference issues is a key element in any system using the radio frequency (RF) spectrum. This is especially true for passive sensing systems—those that do not transmit but only receive naturally occurring emissions—due to the level of sensitivity required for these systems to extract useful environmental and scientific data. Interference can be caused by a variety of sources: other valid users of the RF spectrum, improperly functioning consumer and commercial equipment, and improper or disallowed use of the spectrum. As the use of the RF spectrum for commercial, industrial, government, and scientific uses continues to increase, the number of potentially interfering sources will increase as well. Mitigation techniques are a limited but critical element in efforts aimed at extracting scientific value from an increasingly difficult RF environment.

The Earth Exploration-Satellite Service (EESS) and Radio Astronomy Service (RAS) have classically limited the impact of interference by using mitigation techniques. However, there are physical limits to the capacity of the “unilateral” techniques that typically have been used, and they often do not provide adequate protection from interference. Recently, new techniques have been suggested, in which the active and passive users of the RF spectrum collaborate in order to share the spectrum. These “cooperative” mitigation techniques may provide a potential for meeting the expanding spectral needs of the passive-sensing community.

This chapter is divided into five sections: §4.1 addresses the expected trends in RF spectrum use that may call for increasing mitigation; §4.2, the drivers of spectrum use; §4.3, the capacity for unilateral mitigation technology; §4.4, the potential for mitigation through cooperative use; and §4.5, the costs of mitigation.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 135
4 Technology and Opportunities for the Mitigation of Radio Frequency Interference The capacity to address interference issues is a key element in any system using the radio frequency (RF) spectrum. This is especially true for passive sens- ing systems—those that do not transmit but only receive naturally occurring emissions—due to the level of sensitivity required for these systems to extract useful environmental and scientific data. Interference can be caused by a variety of sources: other valid users of the RF spectrum, improperly functioning consumer and commercial equipment, and improper or disallowed use of the spectrum. As the use of the RF spectrum for commercial, industrial, government, and scientific uses continues to increase, the number of potentially interfering sources will increase as well. Mitigation techniques are a limited but critical element in efforts aimed at extracting scientific value from an increasingly difficult RF environment. The Earth Exploration-Satellite Service (EESS) and Radio Astronomy Ser- vice (RAS) have classically limited the impact of interference by using mitiga- tion techniques. However, there are physical limits to the capacity of the “unilateral” techniques that typically have been used, and they often do not provide adequate protection from interference. Recently, new techniques have been suggested, in which the active and passive users of the RF spectrum collaborate in order to share the spectrum. These “cooperative” mitigation techniques may provide a potential for meeting the expanding spectral needs of the passive-sensing community. This chapter is divided into five sections: §4.1 addresses the expected trends in RF spectrum use that may call for increasing mitigation; §4.2, the drivers of spec- trum use; §4.3, the capacity for unilateral mitigation technology; §4.4, the potential for mitigation through cooperative use; and §4.5, the costs of mitigation. 

OCR for page 135
sPectrum management science 21st century  for in the 4.1 TRENDS IN ACTIVE SPECTRUM USAGE One of the primary concerns for passive sensing systems is the explosive growth in industrial, commercial, and consumer devices. This growth is fueled by user demand, investment capital, and the reallocation of underutilized spectral bands. The need for mitigation and the appropriate mitigation technique will vary depending on the type of equipment that will be permitted by the regulatory agen- cies, the technology being deployed, the time line for the deployment of systems, and the intensity of spectrum usage. This section and the next present a review of current spectrum usage and of the drivers for future spectrum usage, which provides the requisite basis for the development of the appropriate technical and regulatory mitigation strategies. Current Allocations Access to spectrum in the United States is assigned by the Federal Communica- tions Commission (FCC) and the National Telecommunications and Information Administration (NTIA). The process is described in useful detail in Chapter 1 of the National Research Council’s 2007 report Handbook of Frequency Allocations and Spectrum Protection for Scientific Uses.1 To summarize, spectrum is typically assigned to serices (classes of users) on a primary basis or a secondary basis, and allocations include details on permitted transmission power levels and operation times. The difference between a primary allocation and a secondary allocation is essentially that the users of a secondary allocation must accept interference from the users of a primary allocation and conversely must not interfere with the users of the primary service. The International Telecommunication Union (ITU), an agency of the United Nations, periodically updates its allocation table to coordinate international spectrum usage and prevent problems due to interference. The ITU Radio Regulations (ITU-RR) are not binding on the United States in toto—the real treaty obligation of the U.S. government is that it will not assign transmitter licenses in such a way that will cause interference to stations licensed by other governments that are in accordance with the ITU-RR. Within this framework, national govern- ments create and enforce additional regulations, typically to include additional details and to elaborate on permitted uses of the spectrum. In the United States, federal use of spectrum is managed by the NTIA, whereas nonfederal (i.e., com- mercial, amateur, and passive scientific) use of spectrum is managed by the FCC. The authority of the FCC and NTIA are parallel in this respect. FCC regulations 1 National Research Council, Handbook of Frequency Allocations and Spectrum Protection for Scientific Uses, Washington, D.C.: The National Academies Press, 2007.

OCR for page 135
technolo gy o P P o rt u n i t i e s m i t i g at i o n radio frequency interference  and for the of concerning the use of the spectrum are codified in Title 47 of the Code of Federal Regulations (47 C.F.R.). The radio astronomy community is represented in this process as the “Radio Astronomy Service,” and the Earth remote sensing community is represented in this process as the “Earth Exploration-Satellite Service.” A useful synopsis of 47 C.F.R. in terms relevant to the RAS and EESS, including tables of relevant spectral allo- cations, is given in the Handbook.2 For example, this reference text shows that 2.07 percent of the spectrum below 3 GHz is allocated to the RAS and EESS on a primary basis and that 4.08 percent is allocated on a secondary basis (measured in hertz). From a regulatory perspective, the RAS and EESS are comparable to all other services, despite the fact that that they do not transmit. Thus, the allocation of spectrum to the RAS and EESS on a primary basis results nominally (but not actually—see below) in clear spectrum. The allocation of spectrum to the RAS and EESS on a secondary basis is useful mainly in the sense that it offers these services a legal basis for providing input into the use of these allocations. It should also be noted that the allocation of a frequency band to the RAS and/or EESS does not prevent interference even if the allocation is on a primary basis. This is because the effective bandwidth of any transmission is essentially unlimited when observed with a sufficiently sensitive instrument. So, for example, the far out-of-band (side- band) emission of a transmission whose center frequency is properly in a band in which it has a primary allocation may, at some level, appear in nearby bands in which the RAS and/or EESS has a primary allocation. This has historically been a severe problem, particularly with respect to interference from services transmitting from satellites in L-band. (See §3.5 for a discussion of radio frequency interfer- ence [RFI] from Iridium satellites.) In contrast to active uses of the spectrum, the work of RAS and EESS users can be severely affected when the interference power level is far below the internal noise power level of the detection device, since long integration times are usually used in RAS and EESS measurements to reduce the root-mean-square fluctuations in the internal noise. Thus, this issue affects the RAS and EESS in a way that is fundamentally different from the way that it affects active users of the radio spectrum. The spectrum in which the RAS and/or EESS has a primary or secondary allo- cation is relatively small (see Table 4.1). The spectrum in which the RAS and/or EESS has a secondary allocation has diminished usefulness, since there is no pro- tection from the primary users of these bands. As noted in Chapters 2 and 3, the spectrum requirements of the radio astronomy and Earth exploration radio science community currently far exceed the spectrum available to the RAS and EESS on 2 National Research Council, Handbook of Frequency Allocations and Spectrum Protection for Scientific Uses, Washington, D.C.: The National Academies Press, 2007.

OCR for page 135
sPectrum management science 21st century  for in the TABLE 4.1 Total Spectrum Allocated to the Radio Astronomy Service (RAS) and the Earth Exploration-Satellite Service (EESS) Within 9 kHz to 3 GHz Total Bandwidth Allocation (MHz) Percent of Bandwidth Allocated (%) EESS only (9 kHz-3 GHz) Primary 37 1.23 Secondary 122 4.07 RAS only (9 kHz-3 GHz) Primary 62.12 2.07 Secondary 35.5 1.18 RAS + EESS (9 kHz-3 GHz) Primary 62.12 2.07 Secondary 122.5 4.08 NOTE: The percentage and bandwidth allocated to the RAS and EESS between 9 kHz and 3 GHz as of this writing is given in the table above. Note that “RAS + EESS” is much less than the sum of RAS and EESS, particularly in primary bands, showing that the two services are able to share spectrum efficiently. either a primary or a secondary basis. For this reason, these users must routinely observe in bands in which the RAS and/or EESS has neither primary nor second- ary allocations. This is authorized since passive (receive-only) scientific use of the radio spectrum is not prohibited in any part of the electromagnetic spectrum. This is also often technically possible because some parts of the spectrum are sparsely utilized, and services that transmit typically do so with poor spectral efficiency in both frequency and time (although current trends are in the direction of increased usage and improved spectral efficiency; see §4.2). Finding: Owing to their receive-only nature, the passive Earth Exploration-Satellite Service and Radio Astronomy Service, operating from 10 MHz to 3 THz, are incapable of interfering with other services. Finding: Currently, 2.07 percent of the spectrum below 3 GHz is allocated to the RAS and EESS on a primary basis, and 4.08 percent is allocated on a secondary basis (measured in hertz). Current Utilization Studies The allocation of spectrum to a service does not necessarily imply that the allo- cated spectrum is always used for transmission; neither does it imply that the allocated spectrum cannot be used by others for passive scientific observations. In fact, there are a number of ways in which allocated spectrum might remain free of detectable

OCR for page 135
technolo gy o P P o rt u n i t i e s m i t i g at i o n radio frequency interference  and for the of transmissions and available for passive scientific observations. For example, in rural areas with low population density, some services may not be used to any detectable degree; that is, transmissions associated with the active services are typically fewer and weaker in these areas (see further discussion later in this section). If the area is remote enough, transmissions may be sufficiently weak so as not to interfere with radio science observations (although such remote areas are reached by satellite and airborne radio sources). More often, however, the situation is intermediate in the sense that significant interference is observed but can sometimes be managed through a combination of interference mitigation techniques (see §4.3 and §4.4). The “channelization” of frequencies within a given allocation, typically speci- fied either in 47 C.F.R. or as the result of the adoption of an industry standard (e.g., IEEE 802.11), is inherently inefficient. For example, a typical user of the Land Mobile Radio Service might use only a small number of widely spaced channels within the allocated band and may transmit on them only a tiny fraction of the time. Thus even if an active user is received with sufficient strength to prevent the scientific use of some section of the spectrum while that user is transmitting, it is sometimes possible to exploit the sparse “time-frequency” utilization of spectrum by means of the techniques described in §4.3 and §4.4 to observe effectively when the active user is not transmitting. Given the existing trend toward more efficient channelization and increased utilization, however, interference mitigation methods that rely on this property are in danger of becoming less effective over time. The modulation employed by a transmitter may be inherently inefficient, in the sense that it requires a large swath of spectrum but unevenly distributes the power over the channel. An example is the use of the National Television System Committee (NTSC) standard for analog television (TV), which requires a 6 MHz channel but places the vast majority of the transmitted power into just two car- riers constituting only a few hundred kilohertz of bandwidth within this channel (see Figure 4.1). Radio astronomers have been able to observe within active NTSC channels in areas where NTSC transmissions are relatively weak (e.g., deep in the National Radio Quiet Zone [NRQZ]) by observing only within those portions of the channel where relatively little power is present and filtering out those parts of the channel where most of the power is located. However, the introduction of the new digital TV broadcast standard, known as ATSC, makes this technique impossible. This is because the Advanced Television Systems Committee (ATSC) fills the entire 6 MHz channel with a uniform distribution of power, leaving no “hole” through which to observe (see Figure 4.1). The preceding comments can be summarized as follows: (1) The “allocation” of spectrum historically has not implied the “utilization” of spectrum, which has benefited the passive scientific users of the radio spectrum. (2) Technology trends are moving toward more efficient utilization of allocations, in both time and frequency, which is beginning to severely impact the ability to use some bands

OCR for page 135
sPectrum management science 21st century 0 for in the FIGURE 4.1 A comparison of the digital (Channel 10) and analog (Channel 11) television signals. The digital signal is essentially uniform in power across the entire channel, whereas the analog signal transmits most of its information in two narrow bands, leaving holes through which radio astronomers can sometimes observe relatively strong natural sources. Image courtesy of Andrew Clegg, National Science Foundation. Figure 4-1 R01628 uneditable bitmapped image for scientific uses, despite the importance of these uses. As explained later in this chapter, this is true even when taking into account the capabilities of existing and emerging techniques for the mitigation of interference. For these reasons, passive-sensing scientific users of the radio spectrum are greatly concerned with the utilization of spectrum both within the bands allocated to the RAS and EESS and in all other bands accessible to current and planned instruments. Furthermore, radio astronomers are concerned not only with spec- tral occupancy at frequencies at which they wish to observe, but they also monitor transmissions in nearby bands that have the potential to create interference through receiver compression—a condition in which an instrument is desensitized because a signal in a nearby band is present with such great strength that the receiver goes nonlinear. In this case the ability to mitigate the interference through filtering, while

OCR for page 135
technolo gy o P P o rt u n i t i e s m i t i g at i o n radio frequency interference  and for the of retaining sensitivity, is beyond existing technology. The reason for this is that filters must be placed before the saturable active components, and because filters have losses, the system sensitivity is thereby reduced. The largest radio observatories routinely monitor the RF spectrum and typi- cally maintain continuous monitoring programs of some sort. The results of moni- toring campaigns are usually freely available for inspection online—for example, at the interference-monitoring Web sites maintained by the NRAO’s Very Large Array (VLA),3 by NRAO at the Green Bank Telescope,4 and by the National Atmosphere and Ionosphere Center (NAIC) at Arecibo Observatory.5 Unfortunately, these efforts are technically difficult and expensive to maintain and thus often have lim- ited sensitivity and/or restricted time-frequency coverage. As a result, interference that is strong enough to be harmful to radio astronomy may escape detection by existing monitors. With regard to the EESS community, the monitoring of spectral utilization is made even more difficult by the limitations of operations aboard air- craft and satellites and by the coarse spectral resolution of total power radiometers. However, some anecdotal results have been published (see Figures 2.13 through 2.21, 3.10, and 4.2).6 The actual utilization of the radio frequency spectrum has recently become a topic of increasing interest to active users of the spectrum as well. This has resulted in a number of studies reporting measurements of the utilization of the spectrum.7 Typically, the results of such studies report results in terms of spectral occupancy, 3 See “VLA Radio Frequency Interference” at http://www.vla.nrao.edu/cgi-bin/rfi.cgi; accessed Janu- ary 13, 2009. 4 See “Green Bank Interference Protection Group” at http://www.gb.nrao.edu/IPG/; accessed Janu - ary 13, 2009. 5 See National Atmosphere and Ionosphere Center, Arecibo RFI Web site, at http://www.naic. edu/~rfiuser/; accessed January 13, 2009. 6 S.W. Ellingson, G.A. Hampson, and J.T. Johnson, “Characterization of L-Band RFI and Implica - tions for Mitigation Techniques,” Proc. IEEE Geoscience and Remote Sensing Symp. (IGARSS 00), Vol. 3, pp. 1745-1747, July 21-25, 2003. 7 Federal Communications Commission Spectrum Policy Task Force, Report of the Spectrum Effi- ciency Working Group, November 2002, available at http://www.fcc.gov/sptf/reports.html, accessed January 14, 2010; Frank H. Sanders and Vince S. Lawrence, Broadband Spectrum Surey at Dener, Colorado, NTIA Report 95-321, September 1995; Frank H. Sanders, Bradley J. Ramsey, and Vincent S. Lawrence, Broadband Spectrum Surey at San Francisco, CA, NTIA Report 99-367, May-June 1999; S.W. Ellingson, “Spectral Occupancy at VHF: Implications for Frequency-Agile Cognitive Radios,” Proc. IEEE Vehicular Technology Conf. 00 Fall—Dallas, Vol. 2, pp. 1379-1382 (September 2005); A.E.E. Rogers, J.E. Salah, D.L. Smythe, P. Pratap, J.C. Carter, and M. Derome, “Interference Temperature Mea- surements from 70 to 1500 MHz in Suburban and Rural Environments of the Northeast,” Proc. First Int’l Symp. on New Frontiers in Dynamic Spectrum Access Networks (DySPAN 00), November 8-11, 2005, pp. 119-123; Mark A. McHenry and Dan McCloskey, “Multi-Band, Multi-Location Spectrum Occupancy Measurements,” Proc. 00 ISART Conference, Boulder, Colorado (March 2006), available at http://www.its.bldrdoc.gov/pub/ntia-rpt/06-438, accessed January 14, 2010.

OCR for page 135
sPectrum management science 21st century  for in the which can be defined as the fraction of time that a transmission can be detected at a given frequency, for a given sensitivity and a given time-frequency resolution. However, the perception of what constitutes occupancy can be different depending on the measurement and the interests of the person interpreting the results. For example, a recent study performed by the Shared Spectrum Corpora- tion reported 13.1 percent occupancy for New York City and 1 percent at Green Bank, West Virginia, inside the NRQZ.8 By contrast, a study of occupancies in terms somewhat more relevant to radio astronomy applications finds occupancy greater than 30 percent even in the relatively rural areas of Westford, Massachusetts, and Hancock, New Hampshire.9 Both studies are probably internally consistent but cannot be compared because of their different assumptions about the appropriate time-frequency resolutions, thresholds of detection, and tolerable levels of out- of-band (OOB) interference. Measurements are also being made that attempt to bridge this gap by reporting results in terms of cumulative distribution functions (CDFs), which resolve “occupancy” as a function of threshold of detection, and by also quantifying the fragmentation of unoccupied bandwidth. This activity is important, because often in both active and passive uses of the spectrum a mini- mum bandwidth must be available for the channel to be useful.10 While the various efforts of the active and passive user communities have been useful in confirming the sparse time-frequency utilization of the spectrum, most existing studies are of limited help in understanding in detail the potential for interference and for cooperative spectrum use as described later in this chapter. This is due to limited sensitivity (i.e., the inability to detect weak signals that still are sufficiently strong to constitute “occupancy” to a typical user of that band), time resolution that is too coarse to be useful (for example, monitoring a frequency for only a few milliseconds every few seconds, thereby potentially missing strong signals), and frequency resolution that is too coarse to be useful (for example, monitoring bandwidths on the order of hundreds of kilohertz when the signals themselves have bandwidths on the order of kilohertz, thereby desensitizing the measurements) (see Figure 4.2). This is essentially the same problem experienced by the monitoring programs of radio observatories, as mentioned above. Thus, 8 Mark A. McHenry and Dan McCloskey, “Multi-Band, Multi-Location Spectrum Occupancy Mea - surements,” Proc. 00 ISART Conference, Boulder, Colorado (March 2006), available at http://www. its.bldrdoc.gov/pub/ntia-rpt/06-438/; accessed January 14, 2010. 9A.E.E. Rogers, J.E. Salah, D.L. Smythe, P. Pratap, J.C. Carter, and M. Derome, “Interference Tem - perature Measurements from 70 to 1500 MHz in Suburban and Rural Environments of the Northeast,” Proc. First Int’l Symp. on New Frontiers in Dynamic Spectrum Access Networks (DySPAN 00), November 8-11, 2005, pp. 119-123. 10 S.W. Ellingson, “Spectral Occupancy at VHF: Implications for Frequency-Agile Cognitive Radios,” Proceedings of the IEEE Vehicular Technology Conference 00 Fall—Dallas, Vol. 2, pp. 1379-1382 (September 2005).

OCR for page 135
technolo gy o P P o rt u n i t i e s m i t i g at i o n radio frequency interference  and for the of FIGURE 4.2 An example of radio frequency interference measurements made at the Arecibo radio observatory in Puerto Rico on May 26, 2008. The scan from a single location and a single instance in time is from a few megahertz to 1.45 GHz indicating the large number of commercial, government, and consumer uses of the spectrum. Detailed, real-time characterization of the spectrum uses provides Figure 4-2 an opportunity to prevent unauthorized uses of the spectrum from potentially causing catastrophic R01628 interference, as well as the capacity for opportunistically using unused spectrum for enhancing mea - surements. The Arecibo Observatory is part le the National Astronomy and Ionosphere Center, which is uneditab of bitmapped image operated by Cornell University under a cooperative agreement with the National Science Foundation.

OCR for page 135
sPectrum management science 21st century  for in the the passive and active user communities have a common interest in improving the ability to measure the utilization of the radio frequency spectrum. A time resolution of 1 ms would be able to resolve and potentially classify transmit bursts in most mobile radio communications systems using time-division multiplexing (TDM) duplexing or channelization. Such systems use bursts/packets/ frames in lengths of 10 ms to 40 ms due to a trade-off between accuracy in tracking propagation channels and throughput efficiency (payload/header ratio).11 However, a 1 millisecond time resolution would not resolve radar pulses, as these pulses are typically in the 2 to 400 microseconds range.12 Furthermore, if pulses cannot be resolved, it would be more difficult to positively identify the source as radar, as opposed to intermodulation from other things that just happen to be emitting into that frequency. If the pulses are resolved, however, it becomes very easy to identify the source and also to determine whether the sources are “splat- tering,” “jabbering,” or exhibiting other illicit behaviors. For the purposes of the RAS and EESS, and conceivably for many other applications as well, the activity of these radar pulses is of great interest. Even the multipath from these radars can be problematic to sensitive systems.13 To resolve these radar pulses, a time resolution on the order of 1 microsecond would be needed, which would not be technologi- cally difficult to achieve. The bandwidth resolution needed for such a spectrum survey could reasonably be 1 kHz; 1 kHz is roughly an order of magnitude less than the minimum standard bandwidth for any communications system above 30 MHz. A bandwidth resolu- tion of 1 kHz would also resolve most communications below 30 MHz. Also, since much RFI comes in the form of unmodulated carriers (e.g., spurious products from transmitters, stuck microphones, etc.), the bandwidth is lower-bounded only by trans- mitter phase noise and so can be very narrow. A higher bandwidth resolution of, say, 10 kHz, would be too coarse, since most Land Mobile Radio Systems (two-way radios below 1 GHz) are migrating to 6.25 kHz channelization over the next decade. Spatial resolution is the hardest parameter of the space to “saturate” with a monitoring system. Modern cellular systems use cell sizes ranging from building size to tens of kilometers. Satellite- and HAP-based cell systems can have cells hundreds of kilometers in extent. For terrestrial systems, this is highly frequency-, 11 T.S.Rappaport, Wireless Communications: Principles and Practice, 2nd Ed., Prentice Hall, 2002. 12 Frank H. Sanders, “Detection and Measurement of Radar Signals: A Tutorial,” 7th Annual Inter- national Symposium on Advanced Radio Technologies, March 1, 2005; S.W. Ellingson and G.A. Hampson, “Mitigation of Radar Interference in L-Band Radio Astronomy,” Astrophysical Journal Supplement Series, 147: 167-176 (July 2003); G. Miaris, T. Kaifas, Z. Zaharis, D. Babas, E. Vafiadis, T. Samaras, and J.N. Sahalos, “Design of Radiation-Emission Measurements of an Air-Traffic Surveil- lance Radar,” IEEE Antennas and Propagation Magazine, Vol. 45, No. 4 (August 2003). 13 S.W. Ellingson and G.A. Hampson, “Mitigation of Radar Interference in L-Band Radio Astron - omy,” Astrophysical Journal Supplement Series, 147: 167-176 (July 2003).

OCR for page 135
technolo gy o P P o rt u n i t i e s m i t i g at i o n radio frequency interference  and for the of terrain-, and protocol-dependent: different systems have different transmitter den- sities and different typical transmitting powers. Any justifiable angular resolution requirement would be frequency-dependent, such that the survey would achieve lower resolution at lower frequencies and higher resolution at higher frequencies. This relationship has to do with the nature of multipath scattering versus frequency as well as with fundamental limitations in angular resolution—resolution improves with increasing aperture in wavelengths. The ability to locate emitters with suf- ficient accuracy to facilitate the identification of sources would be the goal of the survey, and given the dependencies mentioned above, the necessary spatial resolu- tion would depend on the frequency and on what can be afforded. Finding: Greater efforts to collect and analyze radio emission data are needed to support the enforcement of existing allocations and to support the discussion and planning of spectrum use. Finding: Better utilization of the spectrum and reduced RFI for scientific as well as commercial applications are possible with better knowledge of actual spectrum usage. Progress toward these goals would be made by gathering more information through improved and continuous spectral monitoring. This would be beneficial to both the commercial and the scientific communities. 4.2 MAJOR DRIVERS OF SPECTRUM USE Current measurements of spectral utilization and its impact on passive systems may not be indicative of future spectral use. The drivers for additional spectral bands for intensive use, the allocation of additional bands, and the development of “smart” flexible radio technology will have a profound impact on future use. The following assessment for the time period 2008-2015 is based on well-established drivers, currently allocated spectral bands, and technology that is under develop- ment. This assessment has a high-to-moderate level of confidence. That said, the impact from regulatory changes can be profound—for example, increases in power levels or emission levels permitted outside the primary transmission band could create an RF environment much less useful for passive systems. Assessment of Trends in Spectrum Use for 2008-2015 The current trend toward more intensive use of the RF spectrum will continue unabated for both commercial and government uses. Within the United States, the continued desire for higher levels of access to the Internet (Figure 4.3), coupled with the increased desire for mobility, will incite the development of new commer- cial systems. Technology is also a major driver for more intensive use. New mobile

OCR for page 135
sPectrum management science 21st century  for in the TABLE 4.2 Successes and Limitations of the Unilateral Mitigation Methods Employed to Date by the Earth Exploration-Satellite Service Type of Center Radio Frequency Frequency (GHz)/ Detector Time/ Interference Bandwidth Detector Frequency (RFI) Reference RFI Details (MHz) Type Resolution Pulsed [1] Out-of-band emissions 1.413/20 Kurtosis 36 ms/3 MHz from an ARSR system observed at close range Pulsed [2] Out-of-band emissions 1.413/20 Pulse 10 ns/20 MHz from an ARSR system detection observed at close range Pulsed [3] Unknown source of 1.413/20 Analog 0.5 s/20 MHz presumably out-of- pseudo- band pulses kurtosis Pulsed [5] Airborne flight 1.413/20 Kurtosis 8 ms/20 MHz encountering many source types Narrowband [2] Airborne test flight 5.5-7.7/100 Cross- 26 ms/0.1 MHz encountering many frequency narrowband source types Wideband [2] Airborne test flight 5.5-7.7/100 Cross- 26 ms/0.1 MHz encountering many frequency [4] wideband source types Wideband [5] Airborne test flights 6, 6.4, 6.9, Cross- 26 ms/400 MHz encountering many 7.3/400 frequency wideband source types Gaussian-like None None None None [1] C.S. Ruf, S.M. Gross, and S. Misra, “RFI Detection and Mitigation for Microwave Radiometry with an Agile Digital Detector,” IEEE Transactions on Geoscience and Remote Sensing, 44(3): 694-706 (March 2006). [2] B. Guner, J.T. Johnson, and N. Niamswaun, “Time and Frequency Blanking for Radio Frequency Inter- ference Mitigation in Microwave Radiometry,” I EEE Transactions on Geoscience and Remote Sensing, 45: 3672-3679 (2007). [3] Jeffrey Piepmeier, Priscilla Mohammed, and Joseph Knuble, “A Double Detector for RFI Mitigation in Microwave Radiometers,” IEEE Transactions on Geoscience and Remote Sensing, 46(2): 458-465 (2007).

OCR for page 135
technolo gy o P P o rt u n i t i e s m i t i g at i o n radio frequency interference  and for the of Mitigation Time/ Mitigation Frequency Type Resolution Performance Achieved Comments Frequency 36 ms/ Pulsed RFI ranging between subchannels 3 MHz 1-13 K in 20 MHz detected and removed (post-processing) Time 40 microsec/ Real-time removal of pulsed blanking 20 MHz RFI ranging between 1-20 K in 20 MHz Time 0.5 s/ Pulsed RFI ranging between blanking 20 MHz 1 K and 10-15 K in 20 MHz detected and removed (post-processing) Time 8 ms/ Pulsed RFI ranging from blanking 20 MHz 0.1 to 45 K detected and removed (post-processing) Cross- 26 ms/ Narrowband RFI ranging frequency 0.1 MHz from 1-45 K in 100 MHz detected and removed Cross- 26 ms/ Wideband RFI ranging from Removal of wideband RFI eliminates frequency 0.1 MHz 10-100 K detected and large portions of usable radiometer removed bandwidth; detection possible only when RFI power/MHz substantially exceeds that of noise. Cross- 26 ms/ Wideband RFI ranging from Removal of wideband RFI eliminates frequency 400 MHz ~5-300 K removed large portions of usable radiometer bandwidth; detection possible only when RFI power/MHz substantially exceeds that of noise. None None None Not possible to detect RFI that resembles thermal noise. [4] J.T. Johnson, A.J. Gasiewski, B. Guner, 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, Part 2): 1974-1985 (July 2006). [5] A.J. Gasiewski, M. Klein, A.Yevgrafov, and V. Leuskiy, “Interference Mitigation in Passive Micro- wave Radiometry,” Proceedings of the 2002 International Geoscience and Remote Sensing Symposium, Toronto, Ontario, Canada, June 24-28, 2002.

OCR for page 135
sPectrum management science 21st century  for in the source knowledge are utilized. In the first, the upward-looking measurements of the RAS antenna are augmented with measurements from a “reference antenna” directed toward the source. This latter antenna observes RFI sources at a higher signal-to-noise ratio, which allows better estimation of RFI source properties in the cancellation process. A second approach is utilized for RFI sources with known modulations, for which a demodulation process can increase the signal-to-noise ratio. Given either a demodulation or second antenna measurement, cancellation then involves an estimation and subtraction of RFI source contributions to the data. The latter can be performed either “precorrelation” or “postcorrelation”—that is, either before or after the spatial covariance matrix is formed in an interferometric system. Cancellation performance is limited by the extent to which RFI sources can be detected and successfully estimated (a function of the signal-to-noise ratio at which the RFI sources are observed) as well as by the complexity and any temporal evolution of the RFI environment in which the observations occur. Unilateral Mitigation Successes and Limitations Table 4.2 above provides a short summary of the successes and limitations of the unilateral mitigation methods that have been employed to date by the EESS. Finding: While unilateral radio frequency interference mitigation techniques are a potentially valuable means of facilitating spectrum sharing, they are not a substitute for primary allocated passive spectrum and the enforcement of regulations. 4.4 MITIGATION THROUGH COOPERATIVE SPECTRUM USAGE The unilateral mitigation techniques described in §4.3 are at best a short-term solution to the RFI problem, which can be effective only when spectral occupancy is low and the RFI is easily distinguished from the background. This approach is otherwise inherently limited by the lack of coordination with the active services, and using only this approach, science users would perpetually be “guessing” how to work around RFI. This tactic will soon find its limits given the trends described in §4.1. A far more effective and efficient approach would be bilateral, or coopera- tive, mitigation. Cooperative mitigation techniques would coordinate the timing and regional use of the radio spectrum in a far more dynamic manner than has existed with past technologies and regulatory structures. This is a new approach by which active services cooperate with passive (science) services within shared spectral bands by briefly interrupting or synchronizing radio transmissions to accommodate the science measurements. Such accommodations would occur only when and where those science observations are needed (e.g., during a satellite overpass), so the

OCR for page 135
technolo gy o P P o rt u n i t i e s m i t i g at i o n radio frequency interference  and for the of impact on spectrum availability for active services would be very low. The intel- ligence of modern devices makes this approach attractive, and there will likely be many instances where the costs of this mitigation technology would be readily accepted by users eager to gain access to large portions of the spectrum. Indeed, many devices will need to possess the necessary technologies and standards to negotiate spectrum use automatically among competing users, so the extension of these standards to accommodate science users could, in principle, be accomplished with very low costs and with a very low impact on functionality. Cellular telephones provide a familiar example and some insight as to how this technology could work: cell phone networks automatically coordinate spectrum use among large quantities of transmitting devices. These systems provide a very dynamic command-and-control authority to assign frequency, or to interrupt or deny service, or to give priority (e.g., when a user dials 911) for each device within and among cellular regions. Conceivably, these systems already represent most—if not all—of the needed infrastructure for cooperative mitigation. The only miss- ing elements are the agreed-on standards and the software that would allow these systems to momentarily relinquish assigned spectral bands in response to science requests. These could be communicated either directly from EESS satellites, for example, or from a networked database. Consider the following scenario for cooperative mitigation. For this example, it is assumed that 30 spaceborne microwave radiometers are engaged in Earth observations for operational and research-oriented scientific purposes. This fleet of EESS satellites passes over a specific area several times per day but for only very brief intervals during each satellite’s pass. The typical spot size of an EESS observation on Earth is about 30 km in diameter. Fixed or mobile transmitters operating within or near a receiving band used by the EESS could operate nearly full time if the transmitters were responsive to a blanking request signal or other preprogrammed transmitter time-off period that is coordinated with the over- pass of each EESS sensor. Due to the brief time of footprint passage, this strategy would permit EESS receivers to measure microwave brightness temperatures while negligibly impacting active service performance. This would be especially helpful to EESS observations in bands that are not allocated to the service. To determine the impact on active services, consider the fractional coverage of the fleet of EESS satellites. A “keep-out zone” of 10 times the footprint size, or 300 km in diameter, would generally ensure that the interfering source is well outside the near-sidelobes of the satellite instrument where it is most susceptible to interference. The total keep-out area on Earth for all spaceborne radiometers would then be of order 20π(150)2 = 1.4 million km2, or an area of approximately 0.3 percent of Earth’s total surface. If a random distribution of satellite locations is assumed, this fractional aerial coverage can, to first order, be equated with the fractional probability of occurrence of a satellite observation being made at a

OCR for page 135
sPectrum management science 21st century 0 for in the particular location on Earth. Put another way, an active user could, on average, transmit 99.7 percent of the time within the detection band of a passive satellite sensor without causing any EESS interference. Another cooperative arrangement is illustrated by a hypothetical situation in which all Air Route Surveillance Radars (ARSR), which operate at L-band, would be synchronized to a time standard that allows them to be blanked for the approximately 20 milliseconds for several times per day that the radar transmitter is located within the moving antenna footprint of an overhead EESS sensor oper- ating in L-band. The loss of information to the radar service would be minuscule, and given the ubiquity of modern Global Positioning System (GPS)-based time synchronization and Internet-accessible ephemeris data, the cost of the hardware and software necessary to perform blanking would also be small. However, the value of interference-free data to environmental monitoring and forecasting services at several critical EESS bands, specifically L-band, would be immense. Similar syn- chronization signals could be made available from registered transmitters (both fixed and mobile) to fixed RAS and EESS users or to other users of the spectrum who could then use them to blank their own observations or raise data quality flags. Blanking regions, in certain cases of strong transmitters, could need to be extended to take into account reflections from geographical features. The above arguments and statistics strongly suggest that better time manage- ment of the available spectrum could result in significantly more time-bandwidth product being made available to passive services without impacting active services to any appreciable degree. To simplify the implementation, cooperative strategies are best implemented in bands used by fixed, registered transmitters—rather than unlicensed devices—although most new Internet-connected and GPS, cellular, or Wi-Fi devices could readily be required to contain simple software for cooperative mitigation. Cooperative mitigation techniques have been proposed over the past decade for commercial and consumer devices such as commonly used cordless tele- phones and devices for use in TV white spaces.39 The extension of these techniques to the passive scientific community might provide many benefits. The committee anticipates that the active services could be viable partners in such an arrangement and could benefit from better usage of the active bands as well as from noteworthy public relations through their support of the EESS and RAS. It is conceivable, given appropriate management policy, that the passive spectrum could be “rented” to commercial interests when not needed, with revenues being used to support improved spectral usage studies and/or passive spectrum management needs. Coordination between RAS ground stations and satellites containing trans- 39 P. Kolodzy, M. Marcus, D. DePardo, J.B. Evans, J.A. Roberts, V.R. Petty, and A.M. Wyglinski, “Quantifying the Impact of Unlicensed Devices on Digital TV Receivers,” Technical Report ITTC- FY2007-44910-01, University of Kansas, January 31, 2007.

OCR for page 135
technolo gy o P P o rt u n i t i e s m i t i g at i o n radio frequency interference  and for the of mitters is critical for the present and future viability of the RAS, but it is much more difficult than ground-to-ground coordination, as is discussed in the next two paragraphs. For example, the coordination process between the RAS and Iridium, as discussed in §3.5, shows that coordination is not always successful at reducing RFI to needed levels. As an example of successful collaboration, passive users of the spectrum and the Wireless Medical Telemetry Service (WMTS) were able to find a successful coopera- tive agreement in the 608-614 MHz in which both services still operate. In 1999, the U.S. Committee on Radio Frequencies (CORF) supported the FCC’s proposal for RAS and WMTS to share this band as long as the proposal was enacted in its entirety to include “service rules on eligibility, frequency coordination with RAS facilities, the necessity to protect RAS observations from interference, and technical standards (including field strength, separation distance from the radio observatory, and out-of-band emission limitations).”40 The proposal was enacted as supported by CORF, and the agreement between the services is seen by both parties as an excellent pairing of interests and one that has benefited them both substantially. Similarly, a successful arrangement was made between the Arecibo Observatory and a nearby military radar station. The Puerto Rico Air National Guard operates a frequency-hopping radar with channels between 1220 and 1400 MHz at Punta Salinas, about 75 km from the telescope. Arecibo Observatory staff and the authori- ties at Punta Salinas devised a coexistence arrangement that involves blanking the transmitter when it is aimed at the observatory. This is not meant to say that cooperative mitigation can replace the need for radio quiet areas or for restricted, passive-only bands. Indeed, since the develop- ment of passive techniques often occurs on unscheduled bases and in arbitrary regions, the need for emission restrictions within the small exclusively allocated spectrum and specific geographical zones remains. Many airborne and ground- based EESS experiments require continuous operation within a given zone and would not be able to yield effectively to active systems over time intervals exceeding even a few tens of percents. Such activity requires the use of restricted spectrum. Similarly, for the RAS, transmissions in geographical areas around radio telescopes must be avoided, and in order to maintain existing capabilities it should still be required that the RAS be given a chance to comment on all license applications for fixed and mobile transmitters within prescribed geographical zones around radio telescopes. However (and for example), in a shared time-of-day cooperative scheme, commercial traffic on certain shared bands of RAS frequencies might be 40 National Academy ofSciences, “Comments on Docket No. ET 99-255, Amendment of Parts 2 and 95 of the Commission’s Rules to Create a Wireless Medical Telemetry Service,” filed with the Federal Communications Commission on September 30, 1999. Available at http://sites.nationalacademies. org/BPA/BPA_048830; accessed January 14, 2010.

OCR for page 135
sPectrum management science 21st century  for in the acceptable in exchange for cooperative active access to other bands at suitable times, thus permitting effective radio astronomical observations to take place during transmission-free windows. Finding: Nascent technologies exist for cooperative spectrum usage, but the stan- dards and protocols do not. The above finding is one of the key points of this section: the smart, inexpen- sive, portable, and highly networked electronics that are incorporated into many devices now have the capabilities needed for intelligent spectral sharing, but the organization of the manufacturing sector and the regulatory impetus needed to implement such sharing need to be developed jointly between the scientific and the industrial communities. It is likely that if such coordination can be developed, there will be additional spinoff benefits that will further facilitate spectral sharing within the purely active community as well. 4.5 MITIGATION COSTS, LIMITATIONS, AND BENEFITS As the previous chapters and sections have illustrated, the nature of the costs of the interference problem for the EESS and RAS is wide ranging. The costs are manifested as impaired or even unusable data; costs are also incurred when the EESS and RAS programs must engineer technical or other fixes to mitigate the effects of interference on their operations. Few of these costs can be monetized easily. The reason is that most of the value provided by the EESS and RAS is embodied in public goods—these include the host of environmental benefits and the improved ability to manage natural resources enabled by the EESS, and the enhanced or wholly new science understanding brought by the RAS. By definition, the societal benefit derived from public goods is difficult to express in dollar values. For example, even though improved forecasts are linked to reductions in weather-related loss of life and property, backing out the contribution of EEES data to this outcome is complex and difficult. It is even more complicated to back out from such a calculation the degradation associated with RFI. This very problem is at the heart of spectrum allocation decisions when com- mercial services such as cellular telephones have an easily demonstrated market value, but scientific and other public uses of spectrum do not. As is well known from the literature on the value of public goods, however, simply because these goods are hard to monetize does not lessen their importance to society. Nor does this difficulty reduce the burden on decision makers to accord high importance

OCR for page 135
technolo gy o P P o rt u n i t i e s m i t i g at i o n radio frequency interference  and for the of to public uses in making resource allocation decisions such as those involving spectrum. In this chapter the committee has sought to inform these decisions by highlighting the costs of the interference problem for the EESS and RAS. Earth Exploration-Satellite Service (EESS) The challenge to the EESS below about 10 GHz is from interference arising from high-speed electronics that incidentally radiate isotropically (e.g., electronic cameras and computers), and from short-range wireless services such as Wi-Fi, Bluetooth, and cellular telephones. Interference above about 10 GHz arises from poorly filtered or directed communications, radar, and related services in bands in or near passive bands, or in bands with harmonics in passive bands (see Tables 2.1 and 2.2 and Figure 3.10). Equipment radiating above 10 GHz is mostly sold to large entities at prices well above consumer levels, and mitigating filters or other RFI suppression devices could readily be added to that equipment. One exception is automobile anticollision radar being developed for large-scale consumer sales for use in the 23 GHz band, despite that band’s current worldwide exclusive ITU and FCC passive allocation (see also the discussion in §§2.5, 3.5, and 4.1). A potential future problem could arise if standards for widely used consumer equipment do not preclude incidental emissions above 10 GHz, which generally can be avoided with minor design changes at little cost. Radio Astronomy Service (RAS) The RAS is currently dominated by relatively few large radio observatories located in remote areas that nonetheless are beset by increasing levels of inci- dental interference from proliferating consumer-level electronics such as cellular telephones, Wi-Fi and Bluetooth systems, computers, and so on; from emissions from aircraft and satellites; and from over-the-horizon signals arising hundreds of miles away, well outside most protected areas but reflected by aircraft, the tropo- sphere, and other means. Explicit expenditures for RAS mitigation research and implementation are modest because mitigation for the next generation of radio telescopes will be achieved primarily by the indirect costs of locating the observa- tories in extremely remote locations that are therefore more expensive to develop and operate (e.g., the Western Australian desert or the Chilean Andes). RAS costs are thus arguably already strongly affected by such remote-site mitigation costs, so little mitigation budget is left. Nonetheless, using horizon sensors to detect RFI of terrestrial origin is being pursued.

OCR for page 135
sPectrum management science 21st century  for in the Nature of the Costs of Radio Frequency Interference to the EESS and the RAS The discussion above illustrates that the costs of RFI to the EESS and RAS take several forms. One cost is the direct loss of information when RFI renders data and observations less useful or, in some cases, wholly unusable. This direct loss of information greatly reduces the societal value of the billions of dollars invested in the nation’s EESS and RAS physical infrastructure. Another cost is that of actions that must taken to accommodate RFI, provided accommodation is even possible. As discussed in Chapters 2 and 3, these actions include alterations in sensor deployment and operations, changes in scheduling, and other technical and engineering adjustments. Examples of lost information content include many examples in the cases of both the EESS and RAS. In the case of the EESS these include the following: • In some cases, despite extensive quality checking of EESS data, there are no good means of tracking the impact of a single observation that may be cor- rupted by noise. In the case of radiance assimilation for numerical weather prediction models, a single passive microwave satellite measurement that is contaminated with RFI at a level comparable to the satellite noise is unable to be distinguished from an uncorrupted measurement. The use of such a measurement can cause errors in an entire forecast. • The direct measurement of water vapor and cloud water can be provided only by microwave radiometers. These measurements are commonly made in the 22-24 GHz frequency range, but microwave point-to-point com- munications and automobile anticollision radars operating in this spectral band are a source of significant RFI that will increase as automotive radar becomes more common. • Another example is sea surface temperature, for which measurements are made at 10 GHz. Microwave brightness in littoral regions is impaired by contamination from the use of X-band spectrum adjacent to and within this spectrum allocation. • The 10.7 GHz channels of AMSR-E are RFI contaminated over parts of Europe and Japan and are not used in these locations. (On a research basis, it is still possible to use the 6.9 GHz band for soil moisture retrieval over large regions.) • RFI in bands below 10 GHz can compromise or even render unusable the unique soil moisture information obtained at 1.4 and 6 GHz.

OCR for page 135
technolo gy o P P o rt u n i t i e s m i t i g at i o n radio frequency interference  and for the of In the case of the RAS, examples include the following: • The detection of deuterium formed during the Big Bang and now found in interstellar gas was for many decades impeded by RFI. Detection was possible only after extensive shielding and the use of RFI monitors. • To date, efforts to detect the redshifted (into the VHF-band) 1420 MHz emission of the epoch of reionization have been defeated by RFI; examples include experiments at Arecibo and at the VLA.41,42 • 1612 MHz imaging by the VLA was crippled by legal emissions from the Iridium satellite system until new filters were installed in the VLA. See Figures 3.10 and 3.11. Characterizing these examples of lost information in financial terms is extremely difficult, given the public-good nature of the EESS and RAS. If a loss in the value of the information could be easily monetized, spectrum regulators would have some basis by which to compare the value of spectrum allocations to the EESS and RAS with allocations for consumer products that create many RFI problems. The methodological challenge posed by the comparison of public goods with consumer goods in deciding how best to allocate and manage spectrum among competing uses is well known.43 Another approach to characterizing the costs of RFI involves estimating the costs of the activities undertaken to mitigate or avoid RFI damage. In the case of the RAS, Box 4.1 describes some examples of mitigation costs. By asking what it costs to avoid or mitigate damage, regulators could compare the cost to the EESS and the RAS of avoiding damages with the cost to sources of RFI of mitigating their RFI (such as using filters or other means of RFI suppression). Whichever services face the least cost, either to avoid damage from RFI or to mitigate the creation of RFI, could be asked to bear the financial burden of taking the action. This approach of comparing costs can be useful to spectrum managers. However, because it only looks at costs, and not at benefits to society, of the information provided by the EESS and the RAS, the avoided-cost-based approach is inferior as a means of guid- ing spectrum management. 41 J. Weintroub, P. Horowitz, I.M. Avruch, and B.F. Burke, “A Transit Search for Highly Redshifted HI,” Astronomy Society of the Pacific Conference Series, Vol. 156 (1999). 42 Greenhill, L., et al., “Mapping HI Structures Present During the Epoch of Reionization,” IR&D Report, Center for Astrophysics, Harvard University. 43 Harvey J. Levin, The Inisible Resource: Use and Regulation of the Radio Spectrum, Baltimore: Resources for the Future and Johns Hopkins University Press, 1971.

OCR for page 135
sPectrum management science 21st century  for in the BOX 4.1 Illustrations of Radio Astronomy Actions and Costs for Radio Frequency Interference Mitigation • Example 1: Using knowledge of the local environment. Experts use patterns such as the time of day or year to identify local sources of radio frequency interference (RFI), which can range from a lawn mower to an Iridium-based aerostat used by police for surveillance. In these cases, RFI is solved by coordination. Tracking down the RFI source typically uses about 1 full-time-equivalent (FTE) day to solve. • Example 2: Sleuthing with radio direction finding equipment. This procedure requires an RFI van equipped with receivers, amplifiers, a spectrum analyzer, and a directional antenna—equipment that costs approximately $20,000. In these cases, which are infrequent, RFI is solved by coordination. Here, tracking down RFI may require 2 to 3 FTE days. • Example 3: Tracking ambiguous external RFI. Tracing RFI to a specific satellite source can be time-consuming and difficult. It may also have legal ramifications. These prob- lems can take an FTE month and require archive work, software development, and detailed knowledge of the satellite (system specifications, operating parameters) that may be the source of interference. Summary Increasing levels of incidental interference from proliferating consumer elec- tronics and other sources threaten the EESS and RAS. The primary current EESS problem is active services in passively allocated bands where the atmosphere is sufficiently transparent that EESS instruments see Earth’s surface. The RAS is strongly affected by emissions from aircraft, satellites, and over-the-horizon signals, necessitating the siting of sensitive observatories in remote locations. All RFI poses the potential for loss of information in EESS and RAS observations and data, thus undermining realization of the full societal benefit of Earth and radio astronomy science.