Without any doubt the most innovative of the current crop of proposals for high speed microwave networking, Angel Technologies' HALO system could prove to be a serious competitor to the emerging LEO satellite technology base.

The HALO system is based upon the idea of lifting a communications payload analogous to that carried by a satellite to a stratospheric station above a metropolitan area. The means of lifting the payload is a stratospheric long endurance aircraft, in concept similar to the U-2 of early sixties fame. Instead of pointing their antennas at a satellite in orbit, subscribers point their antennas at a patch of sky above the city, where the aircraft flies in a continuous circular pattern, or orbit.

In effect the HALO system provides a metropolitan area with its own private "pseudo-satellite", sitting at 50,000 ft rather than the hundreds, thousands or tens of thousands of kilometres characteristic of a real satellite.

The idea of an airborne microwave network node is unconventional, but by no means unique. The US Air Force and DARPA are developing a similar concept, called the Airborne Communications Node (ACN), to be carried by the RQ-4A Global Hawk unmanned aircraft, designed originally as a supplement or replacement for the Lockheed U-2 in the reconnaissance role, for which the U-2 is famous. DARPA's ACN payload is currently in development and expected to be flown in the next year or two. If its performance proves adequate, it would be fielded as a supplement to satellites to support air, land and sea based communications systems.

NASA are working on the ERAST program, which is aimed at developing unmanned, long endurance stratospheric aircraft to supplement or replace remote sensing satellites. The use of these aircraft to carry a communications payload is also being explored. The most remarkable product to date of the ERAST program is the massive solar cell powered Pathfinder, and its successor, the 250 ft span Centurion. NASA's intention is to explore the practicality of an unmanned robot aircraft capable of remaining aloft for days and ultimately months at a time, and flying at altitudes as high as 100,000 ft. The typical airliner seldom flies above 40,000 ft. Interestingly, the biggest challenge in the ERAST program has proven to be that of producing an energy storage system of suitable performance, since once the sun goes down there is nothing to feed the solar cells with. The strategy being pursued at this time are fuel cells, which are fed from a storage device which is charged by the solar cells.

There is clearly much activity under way at this time to develop airframes capable of carrying a communications payload into the stratosphere, so the obvious question the uninitiated will ask is why ?

Stratospheric vs Orbital, Aircraft vs Satellites

When Arthur C. Clarke devised the geosynchronous earth orbit (GEO) satellite five decades ago, little did he anticipate how much this technology would change the world around us. Today the planet is ringed by GEO satellites, packed very densely over areas of high population density like the US and Europe.

By the sixties the Russians perfected the polar medium earth orbit (MEO) Molniya satellites, and in the nineties we have Iridium and Teledesic battling over who will own low earth orbit (LEO) and the rest of MEO.

With the satellite market still showing strong growth in broadcast applications and the impending foray by Teledesic into digital interactive applications, it might seem curious that airborne alternatives are being developed. The reasons are however trivially simple to the microwave engineer - latency and path loss.

Latency is the time elapsed between a signal departing a transmit antenna on the earth's surface, propagating up to the satellite, passing through the satellite's hardware, departing the satellite's transmit antenna, propagating back down to earth, and finally impinging upon the earth-bound receive antenna.

For a GEO satellite latency can be as great as 250 milliseconds (or more), for an MEO satellite of the order of 50-100 milliseconds and for an LEO satellite of the order of 10 milliseconds. Such delays are irrelevant if you are broadcasting TV or radio, but they can cause some difficulty with digital data, especially where protocol state machines become involved. Indeed GEO satcom links have been known to cause serious havoc with protocols designed for landlines and unable to accommodate a 1/2 second round trip delay.

Path loss is an equally painful problem with satellites. Disregarding the effects of the lower atmosphere, which we will discuss later, we have to grapple with the reality that the power level into our receiver declines with the inverse square of distance from the transmitter. The further we have to go, the bigger the antennas and the more power we have to push out to achieve a sought bit rate over the channel. Bigger antennas and more power translate directly into costs, at the satellite end and at the ground station end of the link, if we want serious bandwidth. Therefore low cost and high bit rates are mutually exclusive in the satellite game.

Sending microwaves through the lower atmosphere however causes further pain to the satellite system designer, since the lowest layer of the atmosphere, the troposphere, is laden with free water vapour, water droplets as cloud and rain, and gaseous oxygen. If we wish to operate in the less congested centimetric bands, this tropospheric "soup" will soak up our signal very rapidly. Rain and dense clouds are particularly problematic.

If the satellite is directly overhead the distance through this soup is minimal, but if the satellite's position in the sky is moved away from the zenith, things get increasingly worse. This has been one of the big selling points of LEO satellites compared to GEO satellites, in that within the heavily populated temperate zones of the earth your GEO satellite dish is at a disadvantage, operating at a very shallow elevation angle.

If we are bouncing our microwave signals off a payload on a stratospheric aircraft, things change dramatically. The distance between our antenna shrinks from thousands or tens of thousands of kilometers to mere tens of kilometres, with elevation angles typically above 20 degrees. As a result the distance and thus latency drop by a factor of between 25 and 1000 compared to LEO and GEO satellites respectively. No less importantly pathlength loss drops dramatically because of the inverse square law relationship, and long paths through the tropospheric soup are minimised. As a result you can achieve a very high bit rate with a fraction of the transmitted power and antenna size required to achieve the same when bouncing off a satellite.

The physics of microwave propagation and Shannon's law of channel capacity give the airborne system an unbeatable advantage over the satellite.

Aircraft have other, more pragmatic advantages over satellites. They are much cheaper to build and operate, in comparison with satellite manufacture (read hand crafting) and lofting into orbit. Airborne communications payloads can also be designed to a much cheaper commercial avionic standard, since they can be repaired easily on the ground, a choice not available to a satellite operator. The atmosphere shields much of the radiation which can damage satellites and degrade their semiconductor electronics over time thereby contributing to a greater operational life for the electronics.

In practical terms, an airborne package can be maintained in service indefinitely by ongoing repairs, preventive maintenance, and can be upgraded and modified at any time. Those who may doubt this should consider that the B-52s intended to be in service in 2030 were built in the early sixties. Many airliners in service today were built in the fifties and sixties.

Should a satellite be lost due failure, getting a replacement into orbit can take up to years. With aircraft, a spare may be airborne in minutes if it is fuelled and ready.

What are the limitations of aircraft compared to satellites ? The first and foremost, is that they are quite limited in footprint and if we set a 20 degree ground station elevation angle as a limit, can reach out to a distance of about three times their operating altitude. To cover a radius of 100 km the aircraft needs to be at about 100,000 ft altitude. At a 50-60 km radius we get an altitude of about 50,000-60,000 ft, which is the domain in which the HALO and DARPA ACN systems are intended to operate.

The second limitation of aircraft is endurance, unlike a satellite which hangs in orbit, an aircraft needs to be fed on energy to remain aloft. Excluding the NASA ERAST project, this means hundreds or thousands of litres of kerosene to remain up for a decent number of hours.

Addressing the requirements of stratospheric operating altitude and long endurance, such as 8, 12, 16, 24 hours or longer, we end up with a aircraft not unlike the U-2 which is the progenitor of the species - very large wings, and very light structure. Such aircraft are finicky to fly and handle on the ground.

As a result an aircraft capable of replacing a satellite will be limited in terms of the severity of weather it can handle on takeoff or landing, and if 45 knot gusting winds hit its home base airfield, it will have to elsewhere to refuel.

Unlike satellites, aircraft require at this time pilots to fly them, although current trends suggest that unmanned operation may become more commonplace in time. As yet civil air traffic control has yet to catch up with the military in terms of techniques for handling robot aeroplanes in controlled airspace.

Providing an aircraft can be built which has the aerodynamic efficiency to sustain flight for 8-12 hours in the stratosphere with a decently sized communications payload, then the advantages of an airborne system over a satellite can be realised. The HALO system is the first commercial attempt to do so.

HALO and the Proteus Aircraft

The High Altitude Long Operation (HALO) system is currently well into its development phase with the aircraft displayed at the Paris airshow this year, and the communications payload in the design and development phase.

The HALO concept is the brainchild of Dr Nicholas Colella, Chief Technical Officer of the Angel Technologies Corporation startup, and previously involved in US DoD stratospheric drone aircraft projects and portions of the Star Wars scheme. The project is controlled by a consortium including the two major US defence contractors, Wyman Gordon and Raytheon. Raytheon, a major supplier of radar equipment, is developing the communications payload.

The "concept of operations" for HALO is for an aircraft to be continually orbiting on station at a fixed position and altitude above a metropolitan area. The flightpath is envisaged to be a 9-15 km diameter circle. A GPS navigation system will allow the aircraft to maintain its position with an accuracy under 100 metres.

Typically an HALO aircraft would remain aloft for 8 hours, pretty much a limit set by pilot endurance, with three aircraft rotated to maintain a continuous presence. While one is on station another is either climbing or descending to/from station, with the third acting as a hot spare on the ground.

The HALO aircraft will carry a large streamlined 6 metre diameter 0.8 tonne weight communications pod which contains high speed switching equipment and 28 GHz millimetric band transmission equipment and antennas. The pod is designed to pivot while the aircraft banks, and will be stabilised parallel to the surface of the earth. The aircraft can deliver up to 40 kW of DC power to the pod to supply the electronics.

The payload pod is intended to carry up to 100 fixed or independently steerable dish or lens antennas. Two strategies for antenna management are envisaged:

The Proteus "carrier" aircraft for the HALO system was designed and built by Burt Rutan's Scaled Composites in Mojave, the same company which built the record breaking Voyager aircraft which circumnavigated the globe non-stop. The Proteus is largely built with composite materials, and has a gross weight of up to 6.4 tonnes, of which up to 2.8 tonnes is kerosene to fuel the pair of Williams FJ44-2E jet engines. On station endurance is up to 12 hours. Seating is available for two pilots, with an extra seat for a relief pilot or passenger, the cabin is pressurised and high altitude pressure suits are not required. The design station altitude is between 52,000 and 64,000 ft, well above airliner cruise altitudes.

The HALO communications payload is intended to operate in two 300 MHz slices of the 28 GHz band, centred on 27.65 GHz and 28.2 GHz, with a 250 MHz "guard channel" for supporting services at 27.95 GHz. Each of the up to 100 coverage cells would have access to one or more duplex OC-1 channels of 51.84 Mbits/s capacity and 40 MHz bandwidth, which yields an aggregate capacity of the order of about 5 Gbit/s.

Angel technologies claim that a suitable HALO payload could concurrently support between 10,000 and 75,000 1.5 Mbit/s T1 symmetrical channels within a footprint of up to 96 km diameter, making some assumptions about frequency reuse and subscriber load sharing.

The baseline pod architecture described by Angel Technologies is based upon a central high speed digital switch, tied to an array of SONET multiplexers, each of which in turn feeds into a 28 GHz band communications package and antenna.

The subscriber equipment package is relatively simple, comprising a Radio Frequency Unit and a Network Interface Unit. The Radio Frequency Unit contains the steerable dish antenna, intended to be about 0.3 metres in diameter, a millimetric band transmitter of between 100 mW and 1 Watt output power, a millimetric band receiver, and the antenna steering mechanism. Because the aircraft is orbiting at a radius of up to 7.5 km, even at the extremities of coverage the subscriber antenna will have to be steered by a servo loop to maintain track. However, the angular range required is very modest and about several degrees of arc, and the angular rate very slow. Therefore the motion of the antenna will be minimal.

The Network Interface Unit contains the modulators and demodulators for the signal, and the feed to and from the Radio Frequency Unit is done in the 1.4 GHz band, with up/down conversion performed in the Radio Frequency Unit. It is intended that existing 28 GHz terrestrial networking equipment be adapted for this purpose by the addition of antenna tracking hardware and software.

The intent is to provide the user with a duplex 51.84 Mbits/s capacity channel via the Network Interface Unit to a local on site LAN.

Given the use of a central high speed switch in the HALO pod, the system results in a star topology network with a propagation latency between subscribers at the extremities of coverage of about 0.3 milliseconds.

Intended customers for the HALO system include ISPs, corporate users, government users and individual subscribers with a strong demand for bandwidth.

Analysis and Critique

The HALO system has yet to be proven operationally and since it is the first of its kind, there are few parallels against which it can be compared. The idea is radical and innovative, and thus is likely to attract much scepticism if not criticism.

The operational issues in maintaining an aircraft continuously on station are well understood and have been practiced by military aviators for many decades. Plenty of spare parts, spare pilots and ample maintenance personnel can provide for virtually 100% operational uptime. Of course profitability can be impaired to the operator if too much spare hardware and too many spare personnel are required. Given the exceptional service reliability in the airline industry, and the pool of former military pilots in the commercial aviation work-force, this is not a problem area. If an operator has difficulties, it is a problem easily solved.

The technology required for the design and construction of the HALO pod is well established and in many instances very mature. Adapting it for an airborne application will require primarily improvements to the cooling systems to accommodate the very low atmospheric density at the operating altitude. This too is an area which is well understood, since much commercial and military avionic equipment exists which has been designed to operate under such conditions. We can assume that the pod will contain an internal environmental control system (ECS) which will be driven either by the aircraft's generators or bleed air from the engines. Therefore, excluding any engineering bloopers during design, there are no fundamental reasons why this aspect of the system cannot be implemented very robustly.

The ground station equipment is likely to be derived from existing off-the-shelf equipment and the requirement for an additional antenna steering servo loop verges on the technically trivial. Precision antenna tracking techniques were perfected in the sixties and this is not a technologically difficult area.

Ground station visibility of the aircraft will require that antennas be judiciously positioned, with a clear line of sight to the aircraft's orbit. In inner city / CBD areas this will dictate roof mounting. In suburban areas, an exterior wall mount under the roof may be adequate, providing no trees or other foliage obscure the line of sight.

The only genuine criticism I can level at the system design is that the power budgets published by Angel Technologies fall slightly short of what is required to achieve the intended 51.84 Mbits/s channel capacity at maximum range. I simulated the propagation losses using a very recently developed software package which accounted for cloud, rain, oxygen and water vapour losses, and is much more accurate than the standard tables commonly used in satcom design. At 28 GHz oxygen and water vapour absorption can bite quite badly, at modest slant ranges, and rain and dense cloud can be real show stoppers.

The practical consequence of this is that subscribers may need to use antennas of 0.6 to 1 metre diameter, rather than 0.3 metre diameter, to punch through cloud and rain. How frequently such weather conditions arise will depend on the geography of the city in question.

In the context of Australia's underdeveloped and overpriced subscriber level high speed data communications environment, the HALO system looks most interesting. Since most Australian capitals enjoy, by US standards, excellent weather conditions and frequently excellent flying weather, there are no fundamental obstacles to operating such a system in this country with very high availability. Indeed the frequent hot and dry weather over much of the year in many of our capitals would be optimal for the HALO system and possibly allow it to operate over a larger footprint. HALO would cope far worse in monsoonal areas where extreme humidity, dense cloud and heavy rainfall are typical for a large part of the year and a major issue for millimetric band operations.

How the HALO system would fit into our regulatory framework is an interesting point to ponder, as would be the issue of customer acceptance in a market which is frequently shy of new technology.

Should Angel Technologies succeed and establish a successful service in the US, they could become a serious commercial threat to providers such as Teledesic, by displacing them from the lucrative urban and CBD markets. Equally so HALO could damage portions of the emerging ADSL/VDSL markets, by offering a highly capable alternative. Since the basic design lends itself to incremental upgrades, and the use of multiple aircraft to increase coverage footprint and density, it does not suffer the inherently slow ramp up we see in cabled systems.

The perspective view is that airborne networking is about to emerge as a viable technological alternative to satellites, and may produce some interesting long term changes in the market - at the expense of established players.