WEDNESDAY, 6 DECEMBER 2000Dr. Vinton Cerf, one of the designers of TCP/IP, the technical specification behind the Net, begins his talk at the Mill Lane lecture theatre without the technical dullness one might have expected: "Power corrupts, but PowerPoint corrupts absolutely," he remarks on the initial trouble with his presentation. After a fresh burst of laughter from the audience, he gives a surprisingly clear practical analogue of the meaning of the Net. Electric power and motors, the work-horses of the 20th century, whether they operate in refrigerators, lifts, or wherever, have become an integral part of the infrastructure. We barely think of them until they fail. The network of computers is gradually proceeding towards a similar position, processing and distributing information 24 hours a day to make our lives easier.
Arguably the most important starting point of the Net was in 1965 when the late Donald Davies invented packet switching communications. The earlier method is called circuit switching, and it is still being used e.g. in telephone networks. When you make a phone call, physical switches operate to form a dedicated transmission line between your and the receiver's phones. On the contrary, packet switching in its simplest form has a fixed broadcast network between all clients. The information is divided into packets which are labeled by the addresses of the transmitting and receiving terminals. Of all the packets sent to the network, the receiver will accept only the relevant ones and reconstruct the original information.
The ARPANet project started in the USA in 1969 for military purposes. During the cold war, a single mainframe computer was considered too vulnerable, for instance in the case of a missile attack. Hence it was decided to distribute the computing facilities. The network connecting the machines was based on packet switching technology.
As Dr. Cerf notes, contrary to the associations with 'international', the word Internet refers to a network of networks. After the initial ARPANet, by 1973 there were many computer networks in the USA on different communications substrates, such as radio, satellite and cable links. Dr. Cerf was involved in unifying these into something that could work seamlessly, which would require inter-network linkage. Gateways were created to connect the sub-networks, while the IP address system was developed for coherent information transfer between machines on different sub-networks. But just when the importance of IP starts to get a bit too technical, Cerf cunningly reformulates it as "I P on everything" and again we find ourselves laughing.
It took Cerf and his team some five years to finish the TCP/IP protocol which enables modern networking. Contrary to standards handed down from a committee, TCP/IP actually evolved by being used, and it was left sufficiently open to enable further development. It is interesting to notice that the capacity of the Internet backbone then was about 50 kilobits per second, in the order of today's dialup connections and appallingly slow compared to the University network, yet we still use TCP/IP. Not many technologies can claim a similar level of flexibility and openness.
World-Wide Web and commercialization
The most famous application of the Net spawned from CERN, the European particle physics research laboratory. In 1990 the Oxford physicist Tim Berners-Lee came up with a system for the distribution of scientific information and the collaboration of scientist around the globe, based on the Internet. He created HTML from the existing concepts of hypertext, and HTTP (a protocol on top of TCP/IP) for transferring HTML files. This is what we call the World-Wide Web, which I personally think sounds like another name for the Internet, but perhaps HTTP/HTML would not have been so catchy for the public.
Even after ten years, Cerf compares the current situation of WWW to a gold rush. Just as in the traditional analogue, the winners are not necessarily those seeking gold, but often the suppliers of the equipment and infrastructure. Quite a few of the 'dot com' businesses speak for this bitter truth.
Cerf sees a huge growth in the number of users in the future of the net, and not just for the Western elite. A prime example of online enthusiasm is the village of Kihihi, Uganda, where the limitations of the infrastructure are broken through with a satellite link and solar power. It is estimated that there are three billion Internet users by the year 2010. This sets new demands for the existing technology in terms of bandwidth, particularly with the increasing popularity of multimedia services such as real-time audio and video broadcast. The advent of Internet telephony is currently hindered by the packet technology, which may result in delays up to a second, making normal conversation difficult. To reduce this problem, techniques are being developed to increase the priority of voice packets over other data.
Dr. Cerf goes on to explore new applications, some of which are already being used. Most of them appear rather silly without much practical use: speech recognition for controlling your VCR through a home network, Internet-capable refrigerator and a dynamic picture frame for displaying images from the web. However, he reminds us how strange it must have looked when people started using hands-free sets on mobile phones, and seemed to talk by themselves, yet the technology was gradually adopted. For instance, the fridge might be aware of its contents via their bar-codes, and retrieve suitable recipes from the net. It might even remind you by email to your mobile appliance, to buy some milk if the one in the fridge was well past its best before date and about to crawl out. The picture frame would be nice for grandparents, to watch their grandchildren grow, and the images could be updated as often as one likes. Also of interest is the sewing machine by the Japanese Jaguar company, which can download stitch patterns off the net.
The various appliances of wireless Internet access are of particular interest these days. The present GSM technology, while hardly useful for web surfing at 9600 bits per second, is evolving into GPRS (about 100kbps) and 3G/UMTS (about 2Mbps). The two latter utilize packet switching, so the connection can be always on, and network usage will be charged according to the amount of data. However, the current IP address space is limited to 4.3 billion numbers. This will be a problem in the future, when a person may have several IP appliances. The next version (IPv6) will have an address space of about 1038 which, as Cerf rather amusingly pointed out, would suffice even if the Earth's biosphere was completely full of nanoscale devices consisting of a few thousand molecules, each having its own IP address.
The expansion of the Internet to billions of users with increasingly demanding applications poses even further challenges. The ATM technology used in the network backbones is reaching its limits. Reliability and security may be compromised as the result of increasing complexity. Apart from the technical side, there are practical issues with copyrights, liabilities, censorship and the like, which have recently gained media attention with the illegal distribution of MP3 music, for example. In the words of John Gilmore of Sun Microsystems, "The Internet interprets censorship as damage and routes around it." As the Internet as such obeys no rules or borders, common regulations must be agreed on. This will become progressively more difficult in the future, if media like TV and radio are converged into the same network.
Dr. Cerf is currently doing research for an interplanetary network at the Jet Propulsion Laboratory (JPL). Although radio communications have always been essential to space exploration, the aim of the 'InterPlaNet' is to replace the present jumble of communication methods by common standards not unlike TCP/IP. This will reduce the costs, since the same resources can be re-used, and enable the commercial use of space. The latter includes the preparation of certain crystalline materials for semiconductors etc., which will verge on perfection when grown in zero gravity.
The concept of the Internet as the unifying structure between different networks gains special importance in space. The network will be broken up into bundles, which in the largest scale would be those of different planets. JPL expects to have Mars networked by 2010, and a number of other nearby planets between 2020 and 2040.
As we applaud the grand old man of networking, we seem to have gained a whole new perspective on the Net. Looking from outside the hype of WWW, e-this and e-that, it is refreshing to realize that the present situation is but an intermediate stage in the big picture of the evolution of communications. Considering the immense development from the primitive, experimental network of the 1970's to one of the most important media of today, even a modest extrapolation into the future will raise the importance of the Web in our lives into magnitudes hardly imaginable.
- Dr. Cerf's presentations
and other resources
- Internet Society
- Ceiva picture frame
Risto A. Paju is an Undergraduate in Physics at Queens'