|
 By
Ramin Farjad Rad, Ph.D.
Have
you ever wondered how people may have Internet connection while
traveling on board of a ship, or in remote areas lacking any terrestrial
communication and networking infrastructure? Well, I am sure most
people know wireless technology and probably satellites play an
important role in that picture. In case, you might be interested
to know more details of this communication technology, like I did
a while back, I am going to share with you the result of my studies
in this article. I would also like to acknowledge Dr. Farzad Ghazvinian,
the Chief Technology Officer and Senior Vice President of Teledesic
Inc., for the information he shared with me on this topic. Teledesic
Inc., a private company based in Seattle, is planning to launch
the first satellite-based communication network covering the entire
globe in 2005.
Satellite
communication has significantly developed since its original deployment
in early 1960s in the military for voice communication. Although
regulations do not allow the commercial providers to construct and
launch satellites at will, the broad range of applications that
satellite communication can be deployed for has resulted in a rapidly
ramping-up demand for this technology, leading to great technological
advancements in the field and skyrocketing number of satellites
launched every year to the earth’s orbit. In the United States,
NASA has been working with commercial satellite industry to achieve
seamless interoperability of satellite and terrestrial telecommunications
networks. By 1965, more than 100 satellites were being launched
into orbit each year, most of which were merely used for reflecting
radio signals back to earth. During 1970s and 80s, advanced systems,
such as computers and miniature electronics, were designed into
the satellites for more application specific tasks. In 1990s, the
use of satellites grew rapidly for common everyday tasks, and several
companies and development centers have focused on how to exploit
the great potentials of this communication scheme for the public
use.
The growing demand for communication services worldwide, especially
Internet and multimedia, calls for new technologies and architectures
capable of offering high-quality and high-speed connections. In
this context, satellite systems have been identified as a potential
means to meet the explosive Internet demand. A satellite communication
system, having a broad global coverage, bandwidth-on-demand capability,
and the ability to support mobility, is an excellent candidate to
provide Internet services to globally scattered users, especially
those lacking terrestrial Internet infrastructure. As a result,
several new satellite-based technologies and services are emerging,
such as broadband direct Internet access to the home or IP networks
in the space. A simple application of satellite system in providing
network connection is depicted in Figure 1.
This article covers a brief review of the fundamentals and architectures
developed for satellite Internet communications and the integration
of the satellite networks into the global Internet infrastructure.
Satellite Links’ Components and Classifications
The communication process begins at an earth station, which is an
installation designed to transmit and receive high-power high-frequency
(GHz range) signals to and from a satellite in orbit around the
earth. At the satellite station, information is received and retransmitted
to other earth stations within the satellite’s footprint (coverage
area).
A typical ground station consists of a gateway, a network control
center, and a number of operation control centers. The network and
operation control centers perform orbiting control, satellite operation,
and overall network resource management, and the gateway acts as
the network interface between the external networks and the satellite
network while handling addressing and protocol conversations. The
satellite stations are classified into two main categories based
on the orbit they are placed in: geostationary earth orbit (GEO)
and nongeostationary earth orbit (NGEO).
Majority of the satellites in operation nowadays are launched into
the geostationary orbit, which is 35,786km above the earth equator.
A satellite in this orbit revolves fully synchronized with the earth’s
rotation and looks stationary from the earth’s surface. Due
to the limited space available in this single orbit and the large
demand for geostationary satellites, the international regulatory
bodies like ITU (international telecommunication union) have reduced
the satellites minimum spacing to only 2 degrees of longitude. The
high altitude of the geostationary satellites makes them capable
of covering one third of the earth’s surface. However, it introduces
several technical challenges, ranging from significant signal degradation
over the long distance, high transmission power requirements and
hence a need for large antennas, and high cost of launching satellites.
The most important technical difficulty, however, stems from the
relatively long round trip propagation delay (~250ms). This is the
time it takes for the signal to travel from transmitting earth station
to the satellite and back to earth to receiving station. Therefore,
a network round trip delay from transmitter to receiver and back
to transmitter is on the order of 500ms. This long delay makes a
geostationary satellite unattractive for real-time interactive traffic.
Considering the dominant use of applications based on Transmission
Control Protocol and Internet Protocol (TCP/IP) in terrestrial Internet,
a large number of satellites will have to handle TCP/IP-based connections,
although satellite networks themselves are not based on such network
protocols. In a connection using TCP, data throughput is gradually
increased at startup to avoid network congestion, and any data loss
in the link is treated as an indication of link congestion. As a
result, any erroneous data packet in a satellite link causes TCP
to decrease the transmission rate to avoid perceived congestion.
In a GEO link, a connection startup or ramping up the connection
speed after an error detection takes several 500-ms roundtrips leading
to slow and inefficient use of an expensive satellite link.
On the other hand, the nongeostationary orbit’s distance from
the earth surface is from 700km to about 20000km and is divided
into two subgroups: the low earth orbit (LEO) placed from 700km
to 2000km with a round trip delay comparable to the terrestrial
link (less than 50ms), medium earth orbit (MEO) placed at roughly
13,000km and roundtrip delay of ~100ms.
As the footprint of the nongeostationary satellites is typically
much smaller than the geostationary ones, and their relative position
to earth changes continuously, a constellation of fairly large number
of satellites with connection hand-off capability is required for
uninterrupted and global coverage. Figure 2 shows a conceptual view
of the LEO orbit arrangement around the earth, where the LEO satellites
traverse the earth surface in a polar orbit. In these orbits, satellites
pass around any specific point along their path at approximately
7.4Km/s. Therefore, as one LEO moves out of view, another one comes
in at approximately the same time to perform the handoff.
The nongeostationary links require low-power and small antennas
and have shorter roundtrip delays, making them attractive for many
networking applications, especially the access networks and interactive
communication.
The two most common frequency spectra used by today’s satellites
are C-band (4-8GHz) and Ku-band (11-17GHz). The antennas used for
Ku-band are smaller than those of C-band as the antennas size to
receive a minimum signal is proportional to the wavelength, thus
inversely proportional to the frequency. The minimum size for a
C-band antenna is approximately 2-3 meters, while this size can
be reduced to only 45 centimeters for Ku-band. The higher frequency
associated with Ku-band translates into more information carrying
capacity, thus majority of the satellites used for high-speed Internet
connections and video broadcast use Ku-band. Another frequency spectrum
with growing use is Ka-band (20-30GHz), which provides very high
bandwidth and can use very small antennas (~10inches); however,
the environmental impairments (e.g. rain attenuation) degrade the
signal in this band significantly. Due to its high-bandwidth capability,
Ka-band has also received significant attention for high-speed applications.
For example, Microsoft Corp., AT&T Wireless Services, Motorola,
Lockheed Martin, and few investment firms in Saudi Arabia have invested
over $1 billion in Teledesic Inc. that is leading a joint effort
to deploy 288 LEO satellites in Ka-band for a satellite-based network.
Each Teledesic satellite can provide users with 64Mbps downlink
speed and 2Mbps uplink with “fiber-like” quality of service
using only 30-cm dish antennas. The total equipment cost for a Teledesic
terrestrial user is targeted to be less than $1000.
A Network in the Space
In 1993, NASA demonstrated its Advanced Communication Technology
Satellite (ACTS), which was an all-digital, Ka-band, geostationary
satellite system capable of delivering hundreds of Mbps of bandwidth.
Following such a technology milestone, several companies, including
Hughes, Motorola, EchoStar, and Teledesic, received Ka-band licenses
to deliver bandwidths up to 155Mbps to homes and offices. Such broadband
satellite connections can perform a wide range of applications based
on their orbital classification such as high-speed Internet access,
direct TV, video conferencing, email, and interactive distance learning.
Since then, there have been a number of architectural proposals
for satellite-based Internet networks. A satellite network can serve
as part of the Internet backbone, as well as a high-speed access
network (connecting end users to the Internet backbone). Although
the first satellite-based Internet backbone, Atlantic SATNET, was
deployed in early1980s to connect ARPANET in U.S. to the European
research networks, the idea of using satellites for access networks
is fairly new.
The main problem that satellite systems promise to solve is getting
high-bandwidth access to places without a high-speed infrastructure.
Therefore, satellite access networks are not in direct competition
to the digital subscriber line (xDSL) or cable modem in most cases.
The idea of satellites for access networks was further encouraged
by the use of the relatively inexpensive very small aperture terminals
(VSAT), ultra small aperture terminals (USAT) with ~60-cm antennas,
and other development is the communications technology, reducing
the total cost and antenna size of such a communication link. The
VSAT/USAT can deliver up to 24Mbps in a multicast transmission (point
to multi-point) and up to 1.5Mbps in a unicast transmission (point
to point). The introduction of new techniques to deliver on-demand
bandwidth also enables numerous business-on-the-fly applications.
Two main architectures for satellite-based Internet are illustrated
in the following. The first architecture is known as the bent-pipe
architecture, as shown in Figure 3. In the bent-pipe architecture
the satellites act as simple repeaters between two communication
points on the ground without any on-board processing. The satellites
deployed in such a system can be GSO, MEO, or LEO. This link may
be a bi-directional link between the user and the satellite, especially
in the areas that there are no terrestrial connections available,
or can be a high-speed downlink where the uplink is provided by
a slower terrestrial downlink. Even in places with terrestrial connections,
the satellite connection may also be used as a backup link. The
high cost of interactive satellite terminals and the asymmetric
nature of Internet traffic, where the volume of data transferred
downlink is considerably larger than data transferred uplink, has
further encouraged the use of direct broadcast satellites (DBS,
also used for satellite TV). In this scheme, DBS satellite link
is used for high-speed downlink connections, while a cheap terrestrial
link for slow uplink connections. Figure 4 demonstrates such a scheme
where each user has a receive-only satellite dish to collect data
from the high-speed broadcast channel, and a phone line as the reverse
link. Hughes’s DirectPC is an example of such a system.
One drawback of the bent-pipe system is the low spectrum efficiency
and high latency due to no direct connections in the space between
the satellites and therefore requiring multiple links between two
ground points outside a single satellite footprint. In some new
configurations of bent-pipe, on-board processing is used for beam
switching and rerouting to provide efficient channel utilization.
A more complex architecture, which provides a high-capacity communication
and considerably lower latency, is the intersatellite link (ISL),
as shown in Figure 5. While connectivity in space provides higher
capacity and a lot more flexibility, it also introduces complex
routing issues. This architecture uses on-board processing to handle
the two different types of communication between the satellites
in the space and the ground stations. Teledesic Inc. connects all
its 288 LEO satellites using the ISL architecture in Ka-band and
with an aggregate data throughput of 10GB/s. Such a network is designed
to support millions of simultaneous users globally by dynamic channel
allocation to make efficient use of the available bandwidth using
an on-board processor.
…What’s Next
There is a bright prospect that in the very near future satellite
technology, customized for high bandwidth, will provide cost effective
high-quality connectivity and hence Internet access in remote places
on earth. With the extensive amount of developments currently in
effect to reduce the cost and challenges of such a technology, satellites
have a great potential to be one of the major means for global communication.
It can certainly create a major competition to its terrestrial counterparts,
and play a complementary and crucial role in that global picture.
|
