After looking at what signal intelligence (SIGINT) satellites listen to, and having a look at the (partial) history of those satellites in the US, it’s now time for a more technical look at one of those satellites. In 2016, The Intercept published an article about Menwith Hill, a joint UK/US listening post and satellite ground station, using the Snowden documents. Among the published documents, one, dating from 2009, described the satellites soon to be launched, and contained this picture:
It is the first official depiction of a high-altitude US SIGINT satellite. Previously, only informed speculation could be used to get an idea of what those satellites look like, because their existence and capabilities are still classified. Along with the picture came a short text giving the planned launch data of the satellite, it name – ORION -, and its mission – surveying emitters in China then eavesdropping on satellite phones. Using this information, the satellite was identified as USA-202/NROL-26, and Marco Langbroek published an article on its orbital operations.
However, even though the picture has leaked (and has probably been studied to death by the intelligence services of many countries), I did not find any public analysis of its content. So this article is an attempt at one.
In its article on Menwith Hill, The Intercept published another document describing two satellite models controlled from that station: the “mission 7500” MERCURY/MC program, and the “mission 7600” ORION/RIO program. So it seemed like the picture corresponded to mission 7600 satellite. However, this document is not dated, and it is possible it was written before USA-202 was launched. This is supported by the fact it does not mention PAN (USA-207), which was launched shortly after USA-202 in 2009 and was to be controlled from Menwith Hill. So it was possible that USA-202 was something else. Still, it was the best data available.
That changed in August 2017, when The Intercept published another article, this time on Pine Gap, a joint US-Australian satellite ground station and the Australian equivalent of Menwith Hill. Although focused on drone strikes, the article was released with another Snowden document describing the satellites operated by Pine Gap. As for Menwith Hill, the document described mission 7600 satellites, but also their replacement, the mission 8300 satellites. It revealed the first mission 8300 satellite was launched in 2003, and that these satellites provided something special: a second antenna dedicated to electronic intelligence (ELINT). This, coupled with the fact that the document states that these satellites are controlled at two overseas ground stations, and is classified for diffusion to the USA, Australia and the UK, means that Menwith Hill also operates mission 8300 satellites.
The 2009 launch date of USA-202 make it possible that it was a leftover mission 7600 launched late, after the introduction of the first mission 8300 satellites. However, 6 years is a long time for a late launch, and the picture shows clearly a smaller, secondary antenna, as explained in detail below. Since it seems to be a distinguishing feature of mission 8300, USA-202 must be of that type. Consequently, these are its characteristics, taken from the Pine Gap document (bold and italic for emphasis are mine, for an explanation of the acronyms refer to Signal intelligence 101: SIGINT targets):
4.6 Mission 8300 Program Overview
The Mission 8300 system, the earth orbit (GEO) component of IOSA, is a four satellite constellation, and replaces both Mission 7500 and Mission 7600 current systems. Mission 8300 satellites has command and control located at two overseas mission ground sites. The first Mission 8300 spacecraft was launched 9 Sep 03.
4.6.1 System Missions
188.8.131.52 The Mission 8300 system is designed to collect, process, record, and report Signals Intelligence (SIGINT) information.
184.108.40.206 SIGINT support to US military combat operations
220.127.116.11 Crisis monitoring
18.104.22.168 Indications and warning support to the United States and deployed US forces
22.214.171.124 OPELINT operations to collect, identify and geolocate threat emitters (cross-site/cross system operations in conjunction with Mission 8200)
126.96.36.199 COMINT associated with command and control of military forces, movements of VIPs, deployment of military units, states of readiness, training proficiency, and combat operations
188.8.131.52 PROFORMA associated with military data systems (air defense, artillery, etc.)
184.108.40.206 Collection of line-of-sight low/high capacity COMINT signals
220.127.116.11 Collection of communications and electronic signals [Foreign Instrumentation Signals Intelligence] associated with weapons test ranges, science and technology centers, and production and logistics facilities
18.104.22.168 Monitoring testing activity to detect changes in weapons employment doctrine and to verify compliance with strategic arms limitations agreements
22.214.171.124 Collection of satellite/space systems signals
126.96.36.199 Monitor nuclear weapons and high-energy weapons testing [MASINT-Electromagnetic Pulse (EMP)]
4.6.2 Orbital Characteristics
188.8.131.52 The Mission 8300 system has four satellites in earth orbits. Residual satellites may be available to augment the baseline three-spacecraft constellation. It is important to note that the satellites do not have simultaneous access to the entire area where the potential for coverage exists. Actual targeted areas will be determined by specific tasking instructions.
184.108.40.206 Within areas of potential coverage, the effective coverage footprint is determined by the following factors:
1. Primary Factors
a. Emitter frequency: Generally, the targeting of higher frequencies results in smaller collection footprints than for lower frequencies.
b. Look angle: The coverage footprint will increase as the satellite look angle increases, the farther towards the horizon that the satellite looks. Other factors being equal, the footprint will be smallest when the look angle is 0″ (directly below the spacecraft).
2. Other Factors
Emitter transmission power
Emitter antenna type
Emitter antenna orientation
220.127.116.11 Mission 8300 Orbit Benefits:
1. Stable, continuous dwell for 24-hour collection
2. Coverage of primary target areas: Former Soviet Union, China, South Asia, East Asia, Middle East, Eastern Europe, and the Atlantic landmasses.
4.6.3 Operational Characteristics and Capabilities
18.104.22.168 System Characteristics:
1. Four spacecraft in earth orbit
2. Simultaneous multi-mission SIGINT operations (ELINT, COMINT, FISINT and PROFORMA)
3. Common command and control; spacecraft can move between Mission Ground Stations in response to crisis/contingency requirements
4.6.4 Mission 8300 benefits:
1. Increased wideband channels and downlink capacity for wideband collection, especially TECHELINT from modern modulation emitters
2. Dedicated second antenna for ELINT, test range surveillance and signal search
3. High-accuracy OPELINT geolocation capability through cooperative cross-site/cross-system operations with Mission 8200
4. Common high-sensitivity spacecraft design, With capability to relocate spacecraft (orbit nodal repositioning)
5. Collection from different orbit locations through two different MGSs
6. Flexible crisis/contingency support
Armed with this knowledge, let’s see what information we can get from the USA-202 picture.
In order to analyze the picture, I have made an annotated version highlighting the structures of interest:
The annotated feat5ures are:
- The satellite bus
- The ELINT antenna
- One element of the log-periodic antennas
- The main reflector
- A downlink antenna
- Another downlink antenna
- The solar panels
1. Width of the satellite bus. The satellite bus is the yellow, stubby volume with an octagonal base at the bottom of the satellite. It contains station-keeping propellant, as well as the electronics of the satellite. If the satellite is indeed USA 202, then it was launched by a Delta IV Heavy rocket, with a 5-m wide metallic fairing. This fairing has a 4.5-m internal diameter. This means the octagon has to fit in that diameter. Since this octagon is close to a square with slightly tapered edges, this means the bus must have a width around 3.2 m. This gives a scale to the image: all lengths measured in pixel can be converted in meters, for objects parallel to line (1). This works because the picture is CGI and seems to use parallel projection (or a perspective projection very close to parallel), in which lines parallel in the object are parallel in the image, and in which all parallel planes have the same scale.
2. An unfurlable mesh antenna, and a line marking its diameter. This kind of antenna is lightweight, yet can have a large enough diameter to collect relatively faint signals. On the picture, its is probably the secondary antenna dedicated to ELINT collection, test range surveillance and signal search, described in this document. The higher strength of ELINT signals (for instance radars) compared to communications signals probably allows the use of this relatively small antenna compared to the main reflector, and makes it possible to decouple ELINT and COMINT collection tasking, to have a larger footprint for ELINT collection, and to speed up the scanning of large areas (if it is easier to move the smaller ELINT antenna than the main reflector). It seems to be a new addition on the project 8300 satellites compared to the previous-generation ORION. Using the measurement method explained above, the antenna has a respectable 7.2-m diameter.
To better understand why it is there, here are two interesting graphs taken from the work globalsecurity.org did on high-altitude SIGINT. The first describes the power and frequency used by the SIGINT targets that the satellite has to find and listen to:
It can be seen that radars tend to use much higher power than communication emitters. What is not mentioned in the graph is that radars also tend to focus their signals more, so when their scan pattern makes them by chance emit toward a satellite, the satellite receives a lot of signal compared to what it receives from a telecommunication emitter.
The other graph describes the size on the footprint (the region on the ground that the satellite can listen to) depending on the size of the satellite’s antenna and the frequency used by the targets:
This reads as follow: a 10-m diameter antenna (red line) has a footprint with a radius of 5000 km at 100 MHz. For a 7.2-m antenna, it gives a footprint radius of 7000 km at 100 MHz (roughly the frequency of counterstealth air defense radars).
Using a larger antenna would give a smaller footprint, but it would be more sensitive to signals from lower-powered COMINT targets. However for ELINT targets, a smaller antenna seems to be enough and gives a higher footprint to the satellite.
3. Length of an element of what seems to be a log-periodic antenna. The satellite has many of those antennas, mounted on robotic arms. The main feature of a log-periodic antenna is that it can collect signal over a wide range of frequencies, which makes sense to use here, since – as shown in the SIGINT targets graph – the different emitters of interest have very different frequencies. However, as seen above, the higher the frequency, the smaller the footprint, so mounting the antenna on a repositionable arm makes it possible to move the collection footprint on the globe without moving the whole satellite, and to independently task the antennas for simultaneous collection at different points of the globe.
The antennas are pointed upward, whereas the ELINT antenna is pointed downward. This is because the signals do not reach the log-periodic antennas directly: there are first reflected by the main reflector at the top, which collects a lot of signals thanks to its big area, and focuses it.
The length 3. is the one of the longest element of the antennas, and corresponds to the lowest frequency the satellite can collect. The measured length is 2.7 meters, which means the satellite can collect signals with a frequency down to 30 MHz (which corresponds to a 10.8-m wavelength, 4 times the length of the antenna half-element). This is around the lowest a satellite can go, as frequencies under 30 MHz start to be reflected by the atmosphere towards the Earth, and so cannot be collected from space, but can by collected on Earth by listening posts very far from the emitter. It is hard to estimate the highest frequency usable by the satellite, as the smallest elements of the log-periodic antennas are not resolvable on the images. Besides, the ELINT antenna also collects higher-frequency signals and its upper limit is difficult to determine. It is probably imposed by the accuracy of the ELINT mesh antenna’s shape.
4. The signature feature of high-altitude SIGINT is this very large reflector supported by ribs. The signals from the Earth reaching geostationary SIGINT satellites have spread out a lot during the 36 000-km trip, and are thus very faint: they are about a thousand times fainter than if the satellite was in a 1000-km orbit. Furthermore, as the emitters are not cooperative, they are not intentionally pointed at the satellite, contrary to signals sent to communications satellites, so the signals intercepted by the satellite are not the strongest sent from the emitters. Thus a large antenna is needed to collect the faint signals over a wide area, and combine them. Here, because the size required is larger than what would fit in any rocket fairing, the antenna is unfurlable: at launch, it is folded on the white mast visible in the picture, and then deploys.
The ellipse fitted on the border of the antenna in the picture does not have the same shape as the one fitted on the ELINT antenna, which means the reflector is not parallel to the ELINT antenna and the base of the bus, but oriented at an angle, probably like this:
This might be so that the rest of the satellite is not in front of the reflector, and so does not attenuate or interfere with the signals reaching the reflector. Due to this angle, there is an uncertainty on the effective diameter of the antenna. Length 4 is measured at 10.5 m, which would give the antenna a diameter around 21 m. Fitting a perspective projection to the image gives a larger diameter of 29 m, but the fit is not very robust due to the poor resolution of the image. In any case, that is much smaller than the 100-m diameter that is sometimes speculated. However, as the signal collected goes with the square of the antenna size, the main reflector stills collects around 10 times more signal than the ELINT antenna.
Such large deployable antennas are one of the specialties of US satellite antennas manufacturers. In fact, they are the leaders on the commercial market. Harris and Northrop Grumman publicly advertise that they can build very large unfurlable antennas. Harris for instance has built a 12-m diameter antenna for the ICO G1 satellite, and states “Harris offers unfurlable reflector solutions in sizes ranging from 2 to 25 meters”:
Such antennas can have diameters that are 4 times their folded length (for a 2-GHz reflector), so they can be large and still fit in a rocket fairing.
Northrop has built the Astromesh 11-m diameter antenna for the European Alphasat communication satellite, and states “The AM-2 class AstroMesh is designed for very large reflector application, 18-meter to greater than 50-meters”. However, these do not seem to use ribs as on the USA-202 picture. They advertise that their mesh antennas can be used up to the Ka band, so up to around 30 GHz. Northrop also acknowledges one Astromesh is used on a classified satellite.
Interestingly, the Japanese space agency JAXA has launched a satellite in geostationary orbit with two very large 19 m x 17 m reflectors, to test high-speed satellite communications with mobile-phone sized terminals. It is the ETS VIII satellite:
Japanese scientist have worked on a 30-m version of this kind of “Origami” antenna.
5 & 6. Metallic-looking antennas, pointing downward. They are probably the downlink antennas: the signal are collected by the ELINT antenna and the log-periodic ones, amplified, maybe digitized, converted to a higher frequency, and then sent back to the ground stations at Menwith Hill or Pine Gap through these antennas. Although they are small, if they are metallic they can have a very precise shape and so can be used at very high frequencies, such as the Q band. In that case, they can have extremely high bandwidth, necessary to downlink all the wide-band communications intercepted by the satellite, and extremely small footprints on the ground, to prevent interception by foreign powers. Commands from the ground to the satellite might also be received using these antennas, but because little throughput is needed for this, smaller, simpler antennas not coupled to a parabolic dish might also be used.
Dish 5. is not perfectly parallel to the base of the bus, but assuming it is gives the length 5. as roughly 1.5 m. At 50 Ghz, the footprint of the ground would have an 80-km radius, so any intercepting station would have to be close to Pine Gap for instance, and stand out massively in the middle of the Australian desert. However at those frequencies the precise pointing of the dish would become very challenging.
7. A solar panel. It is curved, probably because it is folded over the main antenna in the fairing. It is not very clear if it is attached to the bus (as is usually the case in satellites) or to the main reflector. Since it is not fully in the picture, it is difficult to estimate the maximum power available to the satellite.
Now that we have some idea of how the satellite is made, let’s look at what it actually does. The Pine Gap document shows that mission 8300 satellites intercept communications (COMINT), intercept data exchanges between military systems such as air defense radars and missiles (PROFORMA), monitor weapons tests (FISINT), and detect, identify and geolocate enemy radars (Operational ELINT).
The geolocation capability of pairs of older mission 7600 and 7500 satellites is mentioned in the Menwith Hill satellite descriptions, but the accuracy has been redacted. However, three documents escaped redaction. The first one is an image used in the article, showing the uncertainty ellipse associated to the geolocation of satellite communication antennas:
The image has no scale, but based on the building, the long axis seems to be a few hundred meters long. This photo is also interesting because according to The Intercept it was captured by a US satellite, and indeed the resolution is too good to be from a US commercial satellite, so it is probably from a modern KH-11 satellite.
The other interesting document about geolocation is this one, which mentions the geolocation accuracy of SIGINT satellites against small satellite terminals:
It gives a CEP, basically the size of the error ellipse, of 0.07 nautical mile (130 m). This is coherent with the picture of the ellipse above, and it is a tremendous accuracy for a pair of satellites placed 36 000 km away. Finally, another document mentions a 100-m geolocation accuracy, this time against WiFi signals, so even targets not connected to the internet but using a private network can be located.
A 100-m geolocation accuracy is not enough for direct military targeting in all cases (at least in urban areas, unless you are willing to obliterate all the area), but it is enough to cue other sensors and get a precise fix (as the satellite imagery above shows). It is also enough for direct targeting of emitting vehicles with weapons with a seeker, making the satellite a useful tool in anti-radar operations. These targeting capabilities were mentioned in a 2010 speech by NRO director Bruce Carlson:
The Menwith Hill documents probably apply to pre-mission 8300 systems, so with the new generation, the accuracy figures have probably shrunk a little bit. The quality of the geolocation is so good it could make the Intruder low-orbit SIGINT satellites somewhat redundant, except if they bring something more to the table, like a larger collection footprint.
The picture and the documents published by The Intercept shed light on one of the most secretive aspects of US intelligence, which has been operating since 1968 but which is still largely classified: the high-altitude Signal Intelligence satellites. From their geosynchronous or Molnya orbits, they provide a truly persistent surveillance of the Eastern hemisphere, something optical satellites in low Earth orbit cannot do. However, they cannot collect everything at the same time due to footprint limitations, and their size is smaller than previously thought. This does not prevent them from being an incredibly powerful tool for tactical support to military operations, strategic intelligence and counterterrorism. Although Russia might have launched a geostationary SIGINT satellite in 2014, the USA and its Australian and British spying partners are the only countries possessing such a networked constellation of powerful SIGINT satellites, giving them an edge over other nations, be they adversaries or allies.