|
Services: Back
down to earth
01 April 1999
Teledesic
claims its LEO system will reduce latency and bring
high bandwidth Internet services to those who need
it. But, as Peter Kruger reveals, there's a big
difference between the ballyhoo of announcing a
system and the realities of deploying
it
In
April 1998, when CI looked at satellite-based IP
communication, we identified several important
limitations which restricted the range of Internet
services satellite infrastructure could support.
The foremost limitation of satellitebased Internets
was the latency resulting from a geostationary
satellite's distance from the surface of the earth.
As a packet of data often has to make a round trip
of up to 70,000km before it reaches its
destination, the resulting delay rules out
applications such as videoconferencing and on-line
gaming.
Modifications to satellite based networks, such as
TCP-acknowledge-spoofing, can to some extent
alleviate problems associated with latency.
Spoofing, (which should not be confused with
techniques used by Web hackers), involves the
satellite itself acknowledging packets, rather than
the destination site and Intelsat is one operator
experimenting with spoofing techniques as a means
of reducing latency. However, spoofing methods are
far from being an ideal shortcut, as acknowledgment
of packets isn't just a polite convention - it is
carried out for legitimate and necessary
reasons.
Difficulties would arise if, unknown to the
satellite, the destination site address is no
longer valid. Then, any time saved by spoofing
could be lost, and probably even exceeded, if the
application is unable to cope with being misled
about the validity of its connection. But, as
Stephane Chenard, senior consultant with
Paris-based Euroconsult, explained: "Spoofing is in
the picture - it is better than developing
countless tons of hardware."
Geostationary satellites have large footprints, but
cannot provide global coverage. Operators always
shoot where the ducks are thickest and (pace,
Iridium, ICO Globalstar et al) have left some parts
of the world either very badly served or not served
at all.
As the Internet is global, it is still necessary in
some parts of the world, to route traffic through
landline networks as well as satellites - which
obviates the benefits of using a satellite link in
the first place. If geostationary satellites were
the only type that existed, their use would
scarcely benefit the Internet and would only
provide a means of transferring large amounts of
non time-critical content.
A year ago it seemed consumer applications were
about to become the driving force behind IP over
satellite.
BSkyB, the UK-based satellite broadcaster, which is
part of Rupert Murdoch's News International media
empire, was looking at ways of supplying Web
content to its satellite TV viewers. But, strangely
enough, BSkyB's satellite-based Web project went on
to the back-burner at about the same time as
Murdoch himself went cold on the Internet market,
declaring it overblown and volatile. Erez Gissen,
president of Israel-based IPPlanet, has some
sympathy with BSkyB. "Murdoch is right in believing
that Internet via satellite is not an economically
viable service - and that is because the entire
notion of unicasting via satellite is not a viable
proposition," he told CI. "We believe that
streaming media and push services over satellite,
via multicasting technology, is definitely a viable
service for Internet users."
Someone who claims to be about to change the
satellite industry is Russ Daggatt, vice chairman
of US-based Teledesic. His proposal is to create an
ISP with 288 points of presence (PoPs) orbiting the
earth and providing users with 64Mbit/s downloads
and 2Mbit/s uploads - at a fraction of the latency
of present satellite systems.
Daggatt has his own take on Rupert Murdoch's
attitude to satellite-based Internet. "He is a
content guy," he told CI. "He doesn't own many - if
any - networks. For example, he is the largest
single customer for broadcast satellite services in
the world."
The Teledesic system is a low earth orbit (LEO)
based network of satellites and instead of hanging
around in space 30,000km out from the earth, it
skims the globe at a mere 1,300km. Rather than
routing packets down to the earth's surface, and
passing them on to the user via a landline, data is
passed between adjacent satellites until one is
found which has the destination user in its
footprint. The aim is to provide a radical
reduction in latency and, perhaps, even an
improvement over land-based infrastructure.
"We will average less than 70m for distances of
5,000km or less," claimed Daggatt. "As distance
increases, we will actually gain an advantage over
fibre as signals move through fibre comparatively
slowly - at only two-thirds the speed of
light."
However, selling Teledesic on low latency alone may
be difficult. As Chenard pointed out: "It isn't a
massive problem for a lot of Internet traffic, but
for voice, any latency at all causes big
trouble."
Teledesic, by virtue of the fact that its network
is constantly on the move, should provide the same
level of infrastructure for any point on the globe.
However, the network may struggle in areas where
users, and applications, are bandwidth hungry.
Daggatt doesn't see this as a problem. "Our network
will have high system capacity, but of course, it
is ultimately capacity constrained. However, cost
of service is independent of location."
Teledesic hopes that the availability of high
bandwidth infrastructure will play a role in
improving the economies of developing countries.
However, that could be negated if the service is
not affordable.
Many of the orbiting PoPs will spend years
travelling over areas of the world where there are
not, and will not be, enough Internet-enabled PCs
to generate significant revenue.
Stephane Chenard is reserving judgment on the
pricing of Teledesic. He points out that, despite
claims that the service will be "competitive", the
project itself seems to be in a state of constant
flux, with change after change being made to the
original plan. He commented: "At the moment,
Teledesic can quote what it wants."
The marketing of Internet services will not be
carried out by Teledesic; instead, the company will
appoint resellers and will, if performance meets
expectations, pick up some resellers which are
distributing geostationary satellite service.
However, IPPlanet's Gissen remains to be convinced
about the technical issues. "LEO systems will not
have lower latency on the downlink, because we use
one-hop direct connectivity to the US backbone," he
claimed. But he added, "When LEOs are eventually
available, we will test their performance and we
will use them if it is proven that they can bring
benefits to our customers."
Certainly Teledesic has operators, such as
IPPlanet, in its sights. Daggatt said: "Today, no
service provider can guarantee end-to-end. There
will be many sites for which Teledesic will be the
preferred technology - or, indeed, the only
technology - that can provide the end-to-end
quality of service (QoS) that enterprise
applications demand."
Teledesic is hoping that about 750,000 small to
medium businesses in the US, of which only 3% have
access to fibre, will take up high-bandwidth
services which will run over their network. Daggatt
believes it is these businesses and their demands
for QoS which will give Teledesic the edge. He
said: "Applications will come from the broadband
networks. Just look at where there are broadband
networks today - in the local area networks. These
are the applications that will really drive the
end-to-end QoS requirements."
If the Teledesic system is deployed, (and bearing
in mind the complexity of the project and the state
of the market since Iridium's technical hiatus and
missed targets), it is a big if, then many smaller
operators may be left out in the cold; this is
regardless of whether they are multicasting or
moving large chunks of data via geostationary
satellites.
Daggatt is not concerned about network complexity
issues. "The hand-offs from satellite to satellite
over time are relatively straightforward - it's the
same as a cellular user being handed off from one
cell site to another when driving in a car," he
stated.
Daggatt is also confident about project funding -
and any toll that a fall in Internet-related stocks
may take on Teledesic. He told CI, "As we will be
funding the system over the next four years or so,
it is fair to assume there will be high points and
low points in the financial markets. That said,
stocks of the 'new telcos' like QWEST and Level 3 -
which have funding challenges similar to ours - do
not seem to be linked to the Internet content and
service stocks." So, that's alright then.
Chenard, meanwhile, has doubts about the long-term
commitment of two of Teledesic's investors, Boeing
and Motorola. He pointed out: "Boeing is facing a
severe downturn in its core airline market and
Motorola has other satellite interests -
Iridium."
Above all, he is concerned at the lack of hard
evidence to show that Teledesic is making any real
progress. "We had a big announcement at the
beginning, then there was a conditional approval
from the FCC. Since then, there's been nothing
except a few announcements about personnel
appointments. At this stage, I would at least
expect some more substantive announcements about
suppliers to be made, if nothing else."
He's right. Teledesic does seem to be taking its
time and, as we have seen, the Internet waits
neither for man nor technology. As many rich
incumbent telcos have found to their cost and
chagrin, it makes its own way. Why should it be any
different this time?
|
|
