3. Introduction

Need for the system - Historical Perspective

Year 1969 records the DOD's creation of ARPANET for reaserch in networking. This was also the year when the first RFC was written. Two years after that, the network was connecting 15 nodes, all of them research centers. University of Wisconsin created THEORYNET for internal use, primarily electronic mail.
Then, in 1979 the USENET connected Duke University and UNC. Again research was the main goal of the system.
In 1981, universities that felt left out of ARPANET, join CSNET. In the same year, BITNET is created from the CUNY's cooperative network.
In 1986, NASA and DOE creates NSFNET with backbone speed 56kbps. Next year, computer companies come into the picture, when NSF signed an agreement with IBM, MCI, and Merit Network, Inc. for the NSFNET backbone management.
As technology made progresses, the speed of NSFNET backbone was T1 in 1989.
In 1990, ARPANET becomes history, ceasing to exist because of bureaucracy keeps it from being used to interconnect centers. In 1992, The Internet Society is chartered. The World-Wide Web is released by CERN in the same year.
In 1994 there is an explosion of direct connections to the Internet. Research is not any more the only purpose of Internet.
In 1995, NSFNET reverts back to the research status. The Internet experiences a steady growing process. Not only universities, research companies and other big organizations are part of the Internet. In 1995, companies such as Compuserve, American Online, Prodigy, start to provide on-line dial-up Internet access. The result is that lots of small businesses and homes are now connected. Following is a table showing the growth process summarized above.

Evolution of number of hosts in the period 08/81- 07/96
Date Number of hosts
Aug. 81 213
May 82 235
Aug. 83 562
Oct. 84 1,024
Oct. 85 1,961
Nov. 86 5,089
Dec. 87 28,174
Jul. 88 33,000
Jan. 89 80,000
Oct. 90 313,000
Oct. 91 617,000
Oct. 92 1,136,000
Oct. 93 2,056,000
Oct. 94 3,864,000
Jul. 95 6,642,000
Jul. 96 12,881,000

The estimate for year 2000 is 40 million people connected to the Internet. The large majority will use connections from their homes.

Why Hybrid Technology

Until the technology will be enough advanced such that every family would afford to own and maintain network devices (routers for example) the only choice for establishing a high speed connection from home is the hybrid one.
It is most probably that in the near future homes will have their own network. The modularity of devices and standardization of interfaces will allow interconnection of all household devices. The following figure refers only to the audio/video/data segment of such a home network (the power appliances could also be included).

Fig.3.1 Maximal Home Network
It is clear that for such a system, the incoming flow of information is in most cases higher that the outgoing one. This leads us to the notion of assymetric connection. In this paper we recognize the high demand for high quality audio/video services combined with high-speed Internet access. We will consider the problem of offering integrated network services to users that are connected to the Internet using an Internet Service Provider (ISP). In order to have access to the Internet, a home user must buy a service from a company, such as AOL. The first limitation is the bandwidth available for such a connection. The limitation is of technological nature: maximum rate for analog modems is 28.8 KBps. The typical services that bring the information to such a home network are: hybrid fiber-coax, fiber to the curb, ADSL, DSS, and the newest one hybrid satellite-terrestrial.
The hybrid alternatives are:

Hybrid Fiber-Coaxial Networks (HFC)

HFC is one of the contenders for the 'on ramp to the information superhighway'. Services offered:

Fiber links from the central site to a neighborhood hub, and coax cable from there to a few hundred homes.
Connection requires cable modem to transform the cable signal into a stream of data. However the cable modem is more complex than the analog one: it incorporates other functions like: tuner, network adapters, bridges, router network-management software and encryption devices. Each cable modem has an Ethernet port that connects to the computer on one side and to the cable connection on the other.
The PC is connected directed to the Internet which means that the user need no worry about dialing into the ISP for Internet access.
A typical cable system can serve between 500 and 2,500 homes on a line. Downstream throughput 500kbps - 30 Mbps.
Upstream throughput 96 kbps to 10 Mbps.
Most of the bandwidth is occupied by the TV signals.
The upstream signals travel via a low-frequency band. Thus interferences generated by the household appliances need to be filtered.
Cable amplifiers have to be modified (and in most cases replaced) to separate the two communication ways.
The cable companies need to develop an Internet Point Of Presence for all the networks associated with an head-end.
Today's average price for a cable-modem is $500. A first estimate shows that unit prices will drop to $200 and even $100 in five to 10 years. However, it is the vendors opinion that modem prices won't really drop to $200 or lower until consumer electronics retailers can sell them as standard parts of personal computers .
We also have to consider the investment necessary to set-up the cable infrastructure. For example, to lay fiber for cable TV just in California, the estimate is for $ 15 billion.[Ref2]

Hybrid Satellite-Terrestrial Networks

While there are lots of companies and joint ventures in providing fiber-coax technology, HSTN is only at its beginning.
One of the advantages of these types of networks is that it is much easier to install than HFC. There is no need to alter an existing cabling infrastructure. The user will be provided with an End User Subsystem, for the downstream link, which does not depend on his/her location particularities. The upstream link is already in place as long as the user has a telephone connection.
One of the disadvantages is the unidirectional character of the connection, for the high speed traffic. This disadvantage will be mounted when the technology will be more advanced, to allow high-speed upstream connection.

Both technologies are in the stage when still major changes in structure and/or conceptual planning are being made. While HFC is using an already existing infrastructure, the design of an worldwide HSTN is more challenging.


Abstract To Contents Layout of HSTN