The bandwidth of links to organisations and homes will continue to increase. The averages will only increase gradually.

Many organisations (companies as well as universities) in most advanced countries will expect to have broadband links (better than 10 Mbit/s) to the Information Superhighway backbone.

However, we expect that many homes will at best have modem connections to online services, even in the most advanced countries. Many homes may not make the bandwidth jump up from modem speeds until higher-than-ISDN-speed links (ATM, ADSL, cable modems etc) become available from cable or other providers.

Despite much pushing from government, PTTs and European agencies, broadband services (that is, beyond 10 Mbit/s) will have little penetration outside university campuses and a very few privileged homes and corporate locations in some cities. Services in the region 384 kbit/s to 4 Mbit/s will be available far more widely, in homes and establishments in many cities (this is the ADSL range again).

Thus online services aimed at widespread use in the home sector will have to be delivered in a way which makes them available over a wide range of bandwidth, from modems to broadband. (What does this say about Video on Demand?)

Services aimed at corporate sites (probably even smaller businesses) can assume ISDN speeds as a minimum.

There will be steady but not sharp reductions in telecommunications tariffs over the next 5 years, measured across the board. These tariff drops will be greater in countries moveing towards deregulation; however, in some countries lower telephone tariffs will become available in some cities, especially via cable telephony systems. There will be a gradual move away from time-based tariffs for "telephone-like" calls (including ISDN but perhaps not ADSL); but we expect that time-based charging will still be the dominant mode of charging for the foreseeable future.

Long-distance and international call charges will decline more steeply than local-call charges. Line rental charges could increase as tariffs are "re-balanced".

Tariff differences between countries still are a considerable complication to the network planner. Note that:

Talking of software, the ratios of CPU power, memory and disk capacity per dollar will continue to increase at around the current rate. Thus more and more specialised hardware functions (modems, video codecs etc) will migrate into standard hardware. Telecommunications applications will be limited not by processor or memory restrictions, but by bandwidth availability and cost, and by server costs and power.

One of the constants over the last 20 years has been the survival of the mainframe concept, however disguised. (It is coming back again with a vengeance with the advent of the "Network Computer".) Fully real-time wide-area distributed computing has remained very difficult to achieve at reasonable cost.

The use of TCP/IP seems unstoppable. Worries about IP numbers running out are being resolved by more care in allocating numbers and by the numbering scheme provided by IPv6. Although there are several areas where TCP/IP is not yet adequate, for example in the area of real-time traffic, work is under way on these, and developers have released radio, telephony and video services over the Internet.

Most Local Area Networks are Ethernet (operating at 10 Mbit/s) - with some Token Ring at 4 or 16 Mbit/s. In terms of incremental developments, there is "switched Ethernet" at 10 Mbit/s, and 100 Mbit/s, which are getting very popular with many companies. These technologies do not handle voice or video at all well, but nobody seems to care deeply about that.

For higher-speed networks, and the handling of voice and video as well as data, the technology of choice is ATM. ATM is the strategic choice - most telecommunications operators will in time deploy it, however long a job they seem to be making of it. Nevertheless, it is not yet mature or cheap enough for use on a wide scale in Wide Area Networks.

Overwhelmingly (95%) most corporate and academic networks in Europe are built on a backbone of digital leased lines, mostly at 64 kbit/s and 2 Mbit/s rates, though faster backbones (8 and 34 Mbit/s) are becoming available. ATM technologies are gradually being deployed - oh so gradually when it comes to genuine operational services. How much any of this will affect the home and small business user is not clear, but many would argue that it is largely irrelevant to them for some years still.

Satellite (VSAT) networks are as yet little used in Europe, but have advantages for linking remote areas to the Internet and for one-way data distribution including a high-speed Internet service with terrestrial return channel (the DirecPC concept). Sales of VSAT in Europe are growing, including applications involving small businesses (garages, small shops, etc) - but not homes (not yet, but there are studies on this).

In the last 20 years up to the last few years, access networks were dominated by X.25 technology. X.25 services are dying away fast as they fail to compete in bandwidth or cost terms with Internet providers.

The devices used by the majority of home-based users to access wide-area networks are modems. ISDN cards and adaptors are gradually penetrating this market; and in a few cities, higher-speed cards and adaptors linked to cable or high-speed telephone systems (ADSL or HDSL) are beginning to be used. ISDN adaptors are at last dropping in price and gaining in functionality so that they can operate at speeds of 38.4 or even 57.6 kbit/s and have as good data compression (V.42bis or MNP levels) as modems; however, many users will prefer ISDN cards which operate at the full 64 kbit/s and plug directly into the PC, thus avoiding difficulties with the serial interface. The availablity of 56kbp/s Modems will delay the take-up of ISDN in the home market in some countries, probably including the US and UK.

The Internet is seen by many governments and experts as the basis of the Information Superhighway. However, it is only in the last recently that PTTs and large network corporations think this way. Most PTTs were until recently very cautious about deployment of Internet dial-up services.

Several large network-aware corporations, such as Apple, Microsoft and Novell, were deploying or talking about deploying their own world-wide network for public (or at least corporate) access; virtually all, even Microsoft, have folded their offerings into the Internet.

All agencies will have to come to terms with the Internet as the Information Superhighway as no other new widely deployed rival types of access networks are likely to come into existence.

A number of research groups and companies have developed data broadcasting solutions and deployed them (or talked about deploying them) in Europe. On the whole, most companies have not been enthusiastic about data broadcasting, although in the financial sector (stockbroker) and in racing, they are used.

However, there is a new development in data broadcasting which is much more strategic. This is where data broadcasting is fully integrated with TCP/IP and used as the outgoing channel, with the return channel provided by terrestrial Internet. If one notes that typical WWW browser traffic is highly asymmetric (few keystrokes and mouse clicks upstream, heavy graphics downstream), then one realises that this could be a key application, and a way of bringing "quasi-broadband" Internet to many organisations. Hughes Network Systems have deployed such a service in the US and Europe - several other players including Microsoft have now jumped onto this bandwagon.

In several countries the cable TV situation has been fragmented. Yet cable TV-borne services could be a rival to telephony over copper wire in most European countries. Much though technocrats do not like to admit it, the Superhighway is not dependent on fibre to the home - but it is dependent on a return path from the subscriber and on the ability to switch services. Cable systems can do that, even if they usually need to be upgraded to do it.

Even in the highly deregulated UK the progress has been slow - but better than in most other European countries.

Cable telephony is estimated to have 3.3 million subscribers in 1998 and 4.6 million in 2003. (Up to 30% of newly passed houses in some areas are subscribing to cable telephony.)

It is probable that the cable industry collectively is BT's biggest competitor in the coming years.

The Department of Industry in the UK has awarded over 60 telecommunications licenses in addition to those awarded to cable licensees. Newcomers are able to innovate, "cherry-pick" profitable services and markets, and move on, without affecting large numbers of customers when they change priorities. The push for all kinds of innovation will mean that network operators must search for low-cost operations, while building relationships with service providers, and fostering new kinds of services which can add value to their existing customers. When they [newcomers] are all competing for the attention of a finite customer base a turbulent market must ensue.

There is clear evidence that de-regulation of the UK telecommunications industry is leading to such turbulence. The position of established service providers is being undermined in their most profitable markets, while their regulated requirement to maintain quality services in social markets is eroding profits.

However, the cable industry is insulated from much of this until 2001. It is being offered a unique opportunity to develop competitively in the mass voice market at the core of the telephony industry, while the main competitor (BT) has his hands tied behind his back. Simultaneously they have been granted a monopoly situation in television relay. (Interestingly, and not untypically, technological developments such as ADSL and RealVideo are eroding that regulatory protection.)

Cable is unlikely to fail to achieve its medium/longer term targets in this situation, although there may be disturbances along the way.

Since most European countries have cable networks and as upgrading existing coax systems is now feasible as a short-term measure which can achieve a high penetration of high bandwidth to the home, the move towards a competitive European environment for, in particular, services to the home seems unstoppable. There will be hiccups along the way - in particular it has caused some surprise recently that the EU is no longer forcing the PTTs of Germany or Netherlands to divest themselves of their cable TV interests, despite the obvious (to many) adverse effect on competition.

There are a number of telephony operators in the UK, who are rolling out wireless telephony networks. This is an emerging, turbulent and aggressive market, which requires close scrutiny.

The following points are worthy of observation:

We shall concentrate on Internet Service Providers, since:

1 Internet access is the dominant method for providing access services.

In each country there is a range of Internet providers. These fall into three main types:

It is impossible, and pointless, to keep up to date with all the Internet providers relevant to one's potential needs. There is a great degree of turbulence in the market, and a degree of "shaking out" - surveys written even a year ago are now outdated. International providers are buying up national, and the larger national providers are going international. It is not clear what the long-term future is for local providers but they keep popping up and regrouping.

The best way of getting (temporarily) up to date is to pop into your local newsagent and buy one of the many Internet magazines. You should do this. It is more efficient and probably cheaper than digging the information out from the Web.

Here is a brief list of some Internet providers relevant to the UK:

There are many. Consult your Internet magazine and your local newspapers.

The main technology used for building data networks is still leased digital circuits:

X.25 is a traditional networking service. Every country in Western Europe and most in the East has a public X.25 network. (Some countries have several.) X.25 is not suitable in most circumstances for LAN interconnection (despite a few apologists as the technology is still lurking around). Thus we do not expect that it will be very relevant for most new applications.

This is a much talked about service, oriented to interconnection of LANs. It is mainly deployed at present in the US, and even there on a limited basis. It is on offer in Europe in several countries including the UK.

Some commentators used to predict that Frame Relay would never be comprehensively deployed, since it would be overtaken in the market by ATM. They were wrong, not because Frame Relay is so wonderful, but because commercial Wide Area ATM was so late into deployment.

ISDN is not a very useful service on its own to link LANs, as it is in the nature of most LANs to want to be linked all the time, which does not make sense when you have a time-dependent tariff. However, ISDN is useful to provide backup linking of LANs, and also top-up linking during brief periods of high traffic. It also has a role where LANs need to be linked only occasionally, such as in an office devoted to producing a Sunday newspaper and open only on Saturdays.

The other main problem with ISDN for linking LANs is that the speed of ISDN is too slow at 64 kbit/s. While it is technically possible to bundle up (the technical term is "bond") ISDN calls together, for example to make up a 2 Mbit/s call, the call tariffs go through the roof.

Some telecommunications operators tried to introduce a mid-band ISDN-like service at 384 kbit/s. This all seemed pretty sensible, but it did not catch on.

Above 2 Mbit/s, the leased line options become fewer and more expensive, especially for long-distance and international links. Within cities, things are often much more satisfactory.

However, more and more companies would like such links. So a new breed of network supplier is springing up, stringing fibre from power wires, laying it along railways or in sewers, drowning it in canals, whatever it takes. These suppliers include firms like Energis in the UK, and Veba in Germany, as well as many local cable operators in UK cities.

Above 34 Mbit/s for long-distance links we enter the arena of experimental networks with very limited operational capability. A few rich Internet suppliers have links of that speed from the UK to the US - to serve the whole country's Internet needs!

Commercial services are available in several European countries.

A number of companies now are using the Internet to link their LANs. This can be much cheaper - each site just needs a connection to a nearby Internet Point of Presence (POP) rather than a country- or even continent-spanning leased line connection. The link to the POP is usually a leased line (or something similar like Frame Relay) but is quite short.

Linking LANs via Internet is an attractive concept but it has a number of downsides.

Now that telecommunications is deregulated in many countries, there are a large number of telecommunications operators.

For any country, a good starting point is the relevant PTT. But they are seldom the cheapest operator.

There are three large international consortia that are highly relevant to the planning of international networks:

Regarding national operators, we give just a sample of those operators relevant to the UK:

There are also many local and regional telecommunications operators in the UK, both in London and those based on the regional cable communications companies.

Through the medium of videoconferencing, a lecturer at one site can communicate directly with a group of students located at another site. Students in turn can interact with the lecturer as if (nearly) they were present in the same room as the lecturer.

A typical example of the potential use of videoconferencing is where an educational institution wishes to extend its offering of courses to students located at other institutions some distance away, without incurring the cost of teachers' travel to remote centres.

The advantages of videoconferencing include:

The analysis compares the cost of providing a lecture presented as in a conventional university, to that of presenting the same lecture to students at a remote institution through the medium of videoconferencing. On the conventional model the main cost difference which arises is the additional direct expenditure on travel, and the opportunity cost of the time a lecturer spends in travelling from the home campus, to the remote site. On the other hand, the adoption of videoconferencing incurs additional costs in respect of equipment and telecommunications charges.

The analysis is based on the following assumptions:

(a) Course Teaching:

(b) Student Costs:

(c) Lecturer Costs:

(d) Hardware:

(e) Transmission costs:

Before considering the comparative cost of a face-to-face lecture and a videoconference, the case study will look at the relative costs of 2 Mbit/s and 128 kbit/s circuits. This is because the main analysis can be simplified by eliminating 2 Mbit/s links from the scenario at the outset. The reason for this is that there have been many significant improvements in the field of video compression in the year or two with the result that 128 kbit/s compressed video is now quite acceptable for videoconferencing. Pundits are saying that 2 Mbit/s compressed video is "better", but the cost involved and inflexibility of fixed (as opposed to switched) links is now too much of a penalty to pay.

The chart above shows the cost of installing and using a 2 Mbit/s link. Note that a "kecu" is a kilo-ecu, namely 1000 ecus (about £700). The costs are rather high. The tariff structure is complex. There are fees for the link segments between each site and their local telephone exchanges, and also the trunk segment; plus an annual rental consisting of a fixed payment and a fee related to the length of the trunk segment. It is instructive to spend a moment or two perusing this chart.

The cost of installing the 2 Mbit/s fixed link is distance independent. Spread over five years, it amounts to 7.4 kecu and this is portrayed by the vertical pillar in the foreground which corresponds to a link distance of zero. There is no time-related charge for the link, and this can be observed by looking along the axis labelled video conference (hours); the height of the pillars remains the same, regardless of the number of hours. There is, however a significant distance-related component. Look at any row of pillars parallel to the Link Distance (km) axis, and the height increases as the link distance increases. It can be seen that cost, spread over five years, for a 60 km link is 25 kecu.

The last two sections have described the cost of the operating the 2 Mbit/s fixed and 128 kbit/s dial-up links. The detailed spreadsheets that were used to produce these charts show that the 128 kbit/s is cheaper than 2 Mbit/sec under all circumstances, with only one exception - at zero link distance and 1000 hours of video conferencing.

High speed links using 2 Mbit/s circuits produce high quality video images with none of the jerkiness still associated with the 128 kbit/s circuits. The latter has a complex compression algorithm that seeks to minimise the amount of data that has to be transmitted and the penalty for this is that fast-moving images tend to smear or to jerk. The smearing effect gradually wears away over an interval of several seconds and this process of refining the detail can be disconcerting at first. In practice, people who have used the 128 kbit/s system accommodate to this imperfect imaging very quickly and after a few minutes the effect is not noticed. In summary, the 128 kbit/s system would appear to be adequate for the purpose of distance learning.

The 2 Mbit/s circuits are fixed. Special cable has to be installed between the sites and their local telephone exchanges and this is expensive. The rental is related to the length of the trunk segment and there is no upper limit to the cost incurred this way. Once installed, the circuit is available 24 hours per day but the tariff structure holds no concessions for periods when it is idle.

ISDN circuits have none of the disadvantages that beset 2 Mbit/s circuits. Installation is trivial, no more complicated that installing a conventional telephone line; no special cables have to be laid to the local telephone exchanges. ISDN is a switched service, so it is practical to have a number of videoconference sites across the country or even the EU, all of which can participate in a two-way conference. In fact, it is possible to have several sites participating in a conference by using a multipoint bridge and dedicated codecs. An important observation is that the cost of using ISDN circuits within a country is independent of distance once the distance between sites has entered the trunk call charge band; compare this with 2 Mbit/s, where the cost rises remorselessly with distance. The ISDN service is available 24 hours per day, but the tariff structure is related to the number of hours of use, just like the telephone service.

In summary, the conclusion of the case study must surely be to advocate ISDN circuits for videoconferencing in preference to 2 Mbit/s links, where flexibility and low cost are the dominant forces. 128 kbit/s videoconferencing technology is now mature and it works. Further improvements to the picture quality are in the pipeline, using exotic techniques such as real-time fractal compression.

We will not ask you to do an activity on this material. but if you feel energetic you can check the up to date costings on this situation.