The world is about to be inundated with orbiting mobile comms satellites, if we can believe the marketing hype. The impending deployment of Motorola's Iridium scheme is about to start a new chapter in the communications revolution, one in which we can expect to see mobile communications become quite ubiquitous, worldwide. At this time anything up to a dozen different schemes have been proposed, some of which will succeed and some which may not even get off the drawing board (or CAD system monitor these days).

The first questions which many will ask are, "why now ?", and "so what, we have had satellite comms since the sixties, what's new ?" . In both instances there are very good reasons why. To get a better appreciation of the background to this situation, we need to take a brief look at established satellite communications and their basic attributes.

The Geostationary Earth Orbit Satellite

The Geostationary Earth Orbit (GEO) communications satellite was conceived many decades ago by physicist Arthur C. Clarke, today best known as the author of "2001: A Space Oddysey", who very cleverly figured out that if you hoisted a radio repeater into orbit, you get communications between two points on the surface of the earth which did not have a direct line of sight, and that should that radio repeater be hoisted into an orbit with a rotational period identical to that of the Earth itself, then to an observer on the surface the satellite will appear to hang motionless at one point in the sky. Such orbits are indeed termed "Clarke" orbits, although sadly for the author of the idea, it was not patented and he has not enjoyed his deserved reward for some inspired thought.

Whilst the mechanics of putting a satellite into GEO are quite fascinating, they would clearly exceed the scope of such a discussion. Suffice to say that the satellite is first boosted into a Low Earth Orbit (LEO) parking orbit, from which it is then boosted into a transfer orbit and finally its operating orbit.

Once in orbit the satellite's stabilisation flywheels are spun up, its solar cell clad collectors are unfurled, its position is finely tuned, antennas aligned and it can go to work. Today GEO satellites are a mainstay of global communications, carrying a substantial proportion of the world's voice, data and broadcast TV traffic.

Most GEO satellites in use today are what satellite engineers term "bent pipe" repeaters, which take an incoming radio frequency (RF) carrier wave, heterodyne it to a slightly higher or lower frequency, and beam the signal back to earth. Such devices are needless to say quite devoid of any semblance of intelligence insofar as understanding the content of the data they are carrying goes. As a result, they have very limited functionality outside of carrying multiplexed switched telephone traffic and say broadcast TV video signal relays. While direct broadcast TV has been available for some years now, it too relies upon the GEO "bent pipe" model.

By about the beginning of this decade it was becoming increasingly clear that the "Clarke Belt" about the equator was beginning to become very busy indeed, as the number of physical slots available for new satellites in the most heavily used American and European longitudes was rapidly shrinking, as were the number of available carrier frequency slots. Congestion was indeed setting in, but the market had yet to saturate with demand for the services provided by satellites.

Clearly the satellite had more potential for global communications than had been realised by the GEO paradigm, but more significantly the dual pressures of market demand and unavailability of orbital slots and frequencies meant that other technical alternatives to the GEO would become commercially viable. That and important developments in microwave integrated circuits, which enabled the design of cheap receivers with small, fixed antennas, meant that the time had come for alternatives to the GEO model.

Despite its popularity the GEO satellite does have some important limitations, which are inherent in the fundamental idea. Some of these are particularly important in relation to the transport of computer traffic.

The first and foremost limitation of GEO satellites is round trip latency, which is a measure of the time it takes for a signal to reach the satellite, be turned around and sent, and propagate to the receiver. This time delay, subject to the position of the ground stations, and propagation delays through intermediate repeaters and or encoders/compressors/decompressors and decoders, can be as large as hundreds of milliseconds. It is an inherent property of the GEO geometry and unless somebody devises a warp speed radio wave, it is unavoidable.

Whilst the GEO propagation latency may or may not compromise voice and videoconferencing performance, it has caused difficulties with packet oriented protocols in the past, and was the impetus for the RFC1323 large buffer size protocol revision. Happily many Unix implementations now support the RFC1323 model and no longer experience buffering problems due GEO latency. Promoters of proprietary networking schemes have pushed this as an issue recently, for the record this problem was solved some time ago.

The second limitation of GEO schemes is cost, because the orbit is 36,000 km above the equator, you need to use a big satellite with powerful transmitters, high performance receivers, and large antennas. This satellite, which could weight up to ten tons, then needs to be launched into orbit with a geostationary transfer booster attached to it. The more tons you are pushing out the Earth's gravity well, the more expensive it gets, and the more limited the range of boosters capable of doing the job, further exacerbating the cost problem.

This limitation imposes an essential third limitation, which is that with increasing complexity the potential for hardware failure increases. As a result, designers of GEO sats have been reluctant to include more sophisticated onboard hardware to decode and demultiplex data onboard the satellite. Hence the utter dominance of "bent pipe" designs.

The fourth limitation of GEO sats is their poor performance in the polar regions, as the slant range to the satellite is not only greater, but the radio waves to and from the sat must pass through a very thick layer of the troposphere, the lower 13 km of the atmosphere. Because the troposphere is laden with moisture, it does a very good job of absorbing radio waves. Moreover, since the sat is viewed very low over the horizon, geographical obstacles such as mountains could actually block the view of the satellite.

Other alternatives to the GEO model do exist. One of these is the Soviet devised Molniya (Lightning) orbital model, devised to provide satellite comms and TV broadcasts to remote Siberian sites. The Molniya model uses a clever inclined elliptical orbit which sees the satellite rise gently above the horizon, and then drift down again, significantly reducing the tracking performance needs of a receiver antenna. A constellation of four Molniyas could provide 24 hrs of uninterrupted coverage.

Another alternative are medium altitude constellations of satellites, typically between the destructive charged particle ridden outer and inner Van Allen belts, or placed within the outer Van Allen belt. An MEO constellation has a number of geometrical advantages over the GEO model, because these satellites can cover large areas, have propagation latencies far lower than GEO orbits, and are far less demanding of booster performance. The downside is that they must be more electrically robust to handle the hostile particle environment, and that ground stations must know about the geometry of the orbits in order to find the moving satellites, so they can track them and receive the signal. In decades past, if you wished to receive a signal from such a constellation, you would need a steerable dish on the roof with a clever box of smarts to drive it. Therefore MEO orbits have not been very popular.

The final alternative available to a designer of a mobile satcom network is the Low Earth Orbit (LEO) model, where the satellites are placed into a circular inclined orbit at several hundred kilometres above the surface of the earth. An LEO scheme has the advantages of very low launch costs per satellite, very modest demands upon antenna, receiver and transmitter performance (and thus smaller weight and unit cost), and very low propagation latency times. One limitation of an LEO constellation is the very limited footprint of each satellite, which in turn means that many sats are needed for global coverage. You cannot build an LEO constellation to cover one part of the world alone, it is everything or nothing. Even a modest LEO scheme will require several dozen sats to do the job. Another limitation is the limited life of the sats, which tend to dip into the upper atmosphere eventually and burn up.

LEO and MEO orbital schemes will provide a moving footprint for each satellite "cell", thereby approximating the idea of mobile phone repeaters in orbit, with "fixed" users on the ground. Needless to say some clever design is required to allow user terminals to "hand-over" as one satellite disappears under the horizon and another becomes visible.

The current explosion in planned and proposed comsat schemes is driven by the LEO model. The first scheme to deploy is the Motorola Iridium, the most visible scheme, promoted by Bill Gates, Teledesic, relies on an architecture with several hundred satellites. Grandiose ? Without any doubt the launching of any LEO constellation is a gargantuan pursuit, expected to cost many billions of dollars. One overseas commentator made the apt observation that these schemes rival in scope the building of the pyramids. Not an overstatement by any means.

One remaining and important technical point in these modern satcom schemes is the idea of crosslinks. A crosslink is a satellite to satellite high bandwidth microwave or laser link, which allows a satellite to route traffic to another satellite. Using crosslinks it is possible to wholly bypass the terrestrial fibre and copper networks, a user can beam his message up to a sat over his head, the message can then hop from satellite to satellite until overhead the recipient, from where it is then beamed down to the receiver. This needless to say is a wonderful feature, especially if you wish to cut out the terrestrial-bound competition ! The downside of crosslinks is a significant increase in complexity, as the satellite must carry an onboard switch (or router) as well as steerable crosslink antennas and the associated receivers, transmitters and control hardware (and software).

To gain an idea of the scope of what is being proposed, we will briefly survey some of the most notable schemes.

Inmarsat/ICO

The Inmarsat network of GEO maritime communications satellites is well known, and used for voice comms as well as search and rescue. More recently, Inmarsat proposed the Project21, renamed Inmarsat B and later moved across into a new company, ICO. The ICO scheme involves a constellation of ten MEO sats at 10,355 km altitude, arranged in two 45 degree inclined planes of five satellites, with two orbiting spares. The handsets will be dual mode, defaulting to the orbital link if a terrestrial link is unavailable. The system uses TDMA techniques (like GSM phones), and each sat will support 4,500 phone channels. The satellites, based on the mature Hughes 601 GEO design, weight 2.6 tonnes each, and will operate at 1.6/1.5 Ghz, 1.6/2.4 GHz and around 2 GHz. The ICO scheme is primarily aimed at supporting voice channels.

The ICO scheme should not be confused with the Inmarsat Mini-M scheme, which involves the use of a new generation of GEO satellites, with more powerful beams, and smaller laptop sized user terminals. It is a follow-on to the existing generation of Inmarsat orbital vehicles.

GlobalStar

The proposed GlobalStar system is based upon the model of 48 operational satellites in eight separate LEO planes, at 1414 km altitude and 52 degrees inclination, with eight spares in orbit. GlobalStar uses Qualcomm's CDMA (Code Division Multiple Access) mobile phone technology. The GlobalStar model is tightly integrated with the terrestrial CDMA mobile phone network, and it is assumed that local service providers will licence both to users, as well as deal with regulatory issues. The GlobalStar satellites are essentially bent pipe designs built to relay traffic between ground stations. Each vehicle weighs 450 kg, and uses innovatively, GPS for attitude stabilisation.

Because CDMA allows multiple users to share bandwidth, by using spread spectrum techniques, it has advantages in robustness and lower transmit/receive power over simpler schemes such as TDMA (as used in GSM).

GlobalStar is optimised for voice traffic, but will support low bandwidth data services at 1200, 2400, 4800 and 9600 kbps.

GlobalStar links operate at the following frequencies - terminal to satellite: L-band (1610-1626.5 MHz), satellite to terminal: S-band (2483.5-2500 MHz), gateway to satellite: C-band (5091-5250 MHz), and satellite to gateway: C-band (6875-7055 MHz).

Odyssey

The proposed Odyssey MEO constellation is similar in concept to the ICO scheme, so much so, that Odyssey are now suing ICO for patent infringement. The Oddysey model again employs "bent pipe" repeaters, and frequency division multiplexing of signals. The implementors of Oddysey are TRW Inc. in Redondo Beach, USA, and Montreal-based Teleglobe. AT this time few details appear to be published on this scheme.

Iridium

The first Motorola Iridium is at the time of writing sitting on a Cape Canaveral launch pad waiting for favourable weather and launch window, after an early January abort. Iridium is Motorola's biggest venture to date, and will involve a consortium with a large number of international partners. It will be the first LEO constellation to go operational. The most simple description of Iridium is that of an orbiting GSM mobile phone network (see http://www.iridium.com).

Iridium will employ 66 satellites (originally 77, hence the name) in 778 km high inclined orbits. Each 689 kg satellite is essentially an orbiting telephone switch, with Ka band (~23 GHz) crosslinks to connect to four adjacent satellites in the constellation. Links to user handsets will operate in the the L band (1616-1626.5 MHz), whereas the Ka band (19.4-19.6 GHz for downlinks; 29.1-29.3 GHz for uplinks) si used between the satellite and the gateways and earth terminals. The system uses a combination of FDMA and TDMA techniques, technically conservative, proven but demanding fixed frequency band allocations.

Unlike the preceding designs, Iridium uses a more sophisticated model in which traffic is routed between satellites where terrestrial links are absent. As it is tightly integrated with the GSM protocol suite, it can be made to easily integrate with existing GSM services. This is a tremendous marketing advantage, but carries with it the performance and technical limitations of GSM.

Iridium has provisions for handsets with serial RS-232C interfaces, but due its fundamentally voice traffic oriented circuit switched architecture, it will not be much more effective for data traffic than existing copper networks.

Teledesic

Backed by Bill Gates, Teledesic is the the most ambitious, if not grandiose of the new satcom schemes. In the simplest of terms it could be described as a proprietary broadband orbital competitor to the Internet, or "Internet in the Sky" as one commentator noted, based upon orbital protocol routers.

The Teledesic model envisages no less than 840 (yes, eight hundred and forty) satellites in multiple orbital planes, using a complex scheme of satellite crosslinks to wholly bypass the terrestrial infrastructure. Teledesic is data oriented, and intended to provide services from 16 kbps up to E1 or 2.048 Mbps services. The system is to provide a sustainable usable capacity of 20,000 concurrent 2.048 Mbps connections, with a theoretical limit of 10^6 2.048 Mbps connections, all achieved with extremely low latency times and 10^-9 Bit Error Rates. Teledesic is to employ an ATM based model, with adaptive routing to select the optimal path between satellites. Links are to be encrypted. Provisions are to be made for 155.52 Mbps and 1.2 Gbps uplinks, at selected sites. A complex scheme using FDMA, TDMA and SDMA (Space DMA), plus polarisation control, is to be used to separate channels in adjacent ground cells.

Needless to say, the successful implementation of Teledesic will require some major technological advances in a number of areas. One could say that this is an interesting departure by Gates from his established paradigm of repackaging and dressing up mature technology for sale into the mass consumer market, to a highly technically risky venture which will require often interfacing to highly technical customers. Whether Teledesic is successful and allows Gates to monopolise the world's WAN services, or it becomes a black hole for investors' capital remains to be seen.

Clearly the central focus of most existing satellite mobile comms schemes is voice transmission, with data treated as an ancillary service. With the exception of Teledesic, high performance data services will probably have to wait for the second generation of mobile satellite comms, probably in the latter first decade of the next century.