This commentary is based on Problem 1.5 from my text, Microwave and RF Design of Wireless Systems, John Wiley & Sons: (updated 8/28/00)

The Iridium satellite telephone system consisted of 66 satellites in low Earth orbit (Iridium system poster, photo of Iridium phased array), and was advertised as providing worldwide coverage with a single handset. The system cost was about $5B. The satellite altitude was 483 miles, yielding an expected lifetime of about 5 years, after which the entire constellation would have to be replaced. An advanced feature of the Iridium system was the use of inter-satellite radio links, reducing the number of ground stations, but greatly increasing satellite cost. The handsets were large and bulky, with a typical price tag of about $1000. Service charges ranged from about $1.40 to $3.00 per minute.

While the Iridium system was technologically sophisticated, it was doomed to failure for several reasons. First, the rapid growth of land-based cellular systems provided service to large percentage of the population at rates that typically were a tenth that of Iridium (Typically about $0.35 per minute during peak times, often with free talk time during off-peak hours. Cellular handsets are usually free, or with a small nominal charge). Iridium officials claimed that their system was the only one to offer seamless coverage to people in lesser-developed countries, remote desert or mountainous regions, or even on the oceans. This was true, but they seemed to miss an important point – there are not many paying customers in those regions. Another serious problem with the Iridium system (and one that was never mentioned in their advertisements) is that Iridium handsets required a line-of-sight path to the satellite, meaning that it was rarely possible to use an Iridium phone in a building or vehicle. Land-based cellular telephone systems, working at lower frequencies with better link margins and propagation properties, work quite well in buildings and vehicles. Iridium declared bankruptcy in August 1999, and the present plan is that the satellites will be de-orbited into the oceans. A sad outcome to well-engineered system, but one that was not unexpected.

The Globalstar satellite system consists of 48 LEO satellites (photo of Globalstar satellite), and is also designed to provide worldwide telephone coverage. System cost was about $4B. Globalstar handsets typically cost about $750 (photo of Globalstar handsets), and service charges are about $1 per minute. Satellite altitude is 750 miles, with an expected space vehicle lifetime of about 7.5 years. The Globalstar system does not use inter-satellite relays, but requires many ground support stations. Service began in late 1999, and at the present time the Globalstar system is struggling to meet its market projections. Globalstar requires about 500,000 customers to financially break even, but after six months of operation it had only 13,000. It, too, has trouble providing service to users in buildings or vehicles, and so suffers from the same type of problems as did Iridium. Globalstar's financial backers remain publicly optimistic, but we expect it to suffer the same fate as Iridium, perhaps at a slower pace.

The lesson here is that large constellations of LEO satellites simply cannot compete with land-based systems that provide essentially the same service. Land-based facilities are much cheaper to build, install, and operate than satellites, and they can be much more easily modified, upgraded, and repaired. In addition, the quality of service (including factors such as coverage in buildings and vehicles, handset size, weight, and battery life) of land-based telephone systems is significantly better than that provided by satellite systems. This is ultimately due to the difference in link loss between satellite systems and land-based cellular systems – a fact of nature that no amount of marketing can change. Users will not pay substantially more for inferior service, even if the system can work worldwide. The same conclusion applies to data-oriented LEO systems, such as the proposed Teledesic system.

The Teledesic system is advertised as an “internet in the sky”, and is intended to offer broadband data services (64 Mbps downlink / 2 Mbps uplink). In 1994 Teledesic was proposing to use a constellation of 840 LEO satellites, and had targeted a unit satellite cost of $5.5M (other communications satellites typically range in cost from $50M to $100M). Besides the up and down link hardware, these satellites would also use inter-satellite relays. It has been estimated that this system would require 180,000 millimeter wave phased array antennas, with about 500,000,000 GaAs modules. Anyone with experience in the microwave field will realize the absurdity of these numbers, but the press and pundits like George Gilder have been enthusiastic and uncritical in their praise. (In a 1994 Forbes magazine article comparing proposed LEO satellite systems, Gilder claimed that the “big winner for the next decade is Teledesic”.)

Recently the Teledesic project has been scaled back to 288 satellites, with the first launch in 2000, and service to start in 2004. While there is a huge demand for broadband internet access, this LEO system suffers from the same fundamental problems as Iridium and Globalstar. Competing land-based broadband service can be provided by fiber cable, DSL lines, LMDS and DSS wireless, and others. By the time Teledesic is operational, these services will have picked all the ‘low-hanging fruit’ of the large numbers of customers near urban and suburban areas. And these land based services will offer cheaper rates, with much lower overhead, than the Teledesic system. In the late 1990s, one of the Teledesic technical staff told me that they viewed Teledesic as “a machine that would effectively print money” for the shareholders. We’ll see.

Update July 2000: an engineer from Boeing told me that Teledesic was "dead in the water"

Update March 2001: Iridium assets sold for $25 million.

Update May 2001: Globalstar is close to bankruptcy, with $1.4 billion in debt, and revenues of $1.9 million last quarter