by Brian Vuyk, student at Redeemer University College [1]

Introduction

According to Statistics Canada, as of 2003, 63% of Canadian households were connected to the Internet.1 The Internet has played an increasing role in the life of almost every person, regardless of where they may live. Because of the Internet, it is now possible to easily find information of any sort. The collaboration made possible by the Internet has led to major increases in medical sciences. Inf fact, through modern technologies such as the United Devices Cancer Research Project2 or the Folding@Home Project3, anyone connected to the Internet can contribute towards scientific discovery.

Paul Baran’s ideas had a very large impact on the Internet today. If it was not for his formative ideas concerning redundancy and packet switching, it is quite possible that the Internet as we know it would not exist. However, he is relatively unknown, without the fame of Tim Berners-Lee, creator of the World Wide Web, or Vincent Cerf, who wrote the TCP/IP protocol. As one of the pioneers of network technology, he deserves a more hallowed place among the histories of Computer Science.

A Brief History of the Internet

The Internet dates back to the beginnings of the Cold War. In 1957, the Union of Socialist Soviet Republics has launched the Sputnik satellite. This set off fears in the United States about a possible technological and scientific gap between the two nations. In response to this, before the end of the year, the United States’ Department of Defense (DoD) created the Advanced Research Projects Agency (ARPA) in order to establish an American lead in science and technology. One of the earliest concerns of the DoD was the issue of communications in the event of a nuclear war. The telephone system of the time consisted primarily of switching offices, each connected to thousands of telephones. Each of these switching offices were in turn connected to a larger called toll offices. Each toll office was in turn connected to one or two other toll offices, with little redundancy. Their concern was the the destruction of a few central offices in the event of a nuclear war would fragment the telephone system into many small islands of communication.4

Near the beginning of 1960, the United States Air Force commissioned Paul Baran of RAND Corporation to study the possibility of creating a decentralized network which could survive a nuclear attack, while allowing the United States to retain command and control of communications to it’s Army, Air Force and Navy in order to launch a counter attack. The United States saw a survivable communications framework as a necessity in the case of a nuclear attack, in order that proper command and control might be maintained.5

Baran, by this point, had already independently done some research into the field of “survivable communications”. As an ethical issue, he felt that if a communications system could created which could survive a nuclear strike, the temptation to launch a preemptive strike would be much reduced for military leaders.6

In the early sixties, Baran submitted multiple reports to the Department of Defense describing a redundant network involving multiple backup links between telephone switching offices and multiplexing stations. Baran planned that the switching stations be located far away from population centres, because population centres were considered to be military targets. Due to the redundancy, it would be very difficult for any users to be cut off from the rest of the network.7

One of the initial problems with this network was that due to their remoteness, the paths between switching stations was now much longer than could be traversed by analog signals without considerable distortion. Baran’s solution to this was to use a digital packet switching system throughout the system.8 In addition to allowing greater transmitting distances, this new packet switching design allowed the transmission of much more than voice signals over the network; in his new system, a message could take the form of anything from digitized voice to computer data.

This was accomplished by digitizing all data before it was sent. This was accomplished by sending all data to a multiplexer before it was sent. The multiplexer would convert the incoming signal into a sequence or binary numbers or “bits”. The multiplexer would then divide this bit sequence into a series of smaller blocks 1024 bits long, and attach a header of binary data containing source and destination information. This combination of bits and header was called a packet. Each packet, as it was created, would be sent to the local switching office. This office would examine the header data, and decide on which switching office to send the packet to next. This would continue until the data was delivered to it’s destination. At the destination, another multiplexer would strip the header information from the packets, and reassemble the blocks into their original sequence.

The beauty of this system, was that every packet contained it’s own set of information9, so that should the need arise, a packet could be sent along a different route if needed, and still arrive at it’s destination intact. Each node would decide for itself the best next step for packets coming in. This would prevent the need for a single operations center, which, as Baran noted, “forms a single, very attractive target in the thermonuclear era”10 . This network would allow for survivability and high capacity; the communications possible over this new network would be even greater than those possible under the current telephone and telegraph systems.

Impressed with Baran’s plans for a new network, the DoD began to look around for a company to implement Baran’s network. They approached AT&T, then the largest telephone company in North America, and requested that AT&T build a prototype of Baran’s network. Perhaps feeling threatened by Baran’s work, AT&T dismissed Baran’s work out of hand, and declared that it couldn’t be built. 11

In 1967, Larry Roberts, the director of ARPA, turned the direction of ARPA to networking. At this point, ARPA had funded defense-oriented computerized research programs at multiple universities across the US. It was decided to attempt to build a network, linking each of these projects, in order that the results could be more easily shared12. Mandated with building a large, multi-node network, but with little idea about how to go about building one, Roberts wrote a paper about the possibility of creating a packet switching system subnet.

At the ACM SIGOPS Symposium on Operating System Principles in Gatlinburg Tennessee in late 1967, Roberts presented the vague paper proposing his idea13. Unexpectedly, at the same conference was a paper presented by Roger Scantlebury and Dr. Donald Davies. of the Britain’s National Physics Laboratory (NPL), entitled, “A Digital Communication Network for Computers Giving Rapid Response at Remote Terminals”14. In this paper, Scantlebury and Davies described the design and implementation of a small network using the principles of Baran, along with other principles developed by Davies.

Upon hearing about the work done by Baran and Davies, Roberts had the beginnings of ideas on how to implement his network. Using ideas taken both from Baran and Davies, the ARPANET project was underway. The first step was to contract the construction of the network nodes to the acoustics and computing firm of Bolt, Beranek and Newman (BBN). In 1969, Derek Barber from NPL came to visit the team at BBN. He reported back the the BBN team was interested in the possibility of connecting the NPL network directly into ARPANET.15 In 1967, after reading much of Baran’s work, Roberts hired Baran to join his ARPANET group. Because of this, many of Baran’s original ideas were incorporated into ARPANET.

In December 1969, the first four nodes on ARPANET went online. By March 1971, the infrastructure to connect and additional 11 selected sites to the intranet was in place. The fifteen sites connected to the ARPANET as of March 1971 were 16:

• Stanford University
• Stanford Research Institute
• University of California, Santa Barbara (UCSB)
• University of California, Los Angeles (UCLA)
• University of Utah
• University of Illinois
• Case
• Carnegie Mellon University
• Systems Development Corporation
• Lincoln University
• Massachusetts Institute of Technology
• BBN
• Harvard
• Burroughs

The time was coming to unveil the ARPANET to the world at large. In October, 1972, the First International Conference on Computer Communications (ICCC) was to be held in Washington DC. BBN’s Robert Kahn began to actively urge programmers at different institutions to create new applications or modify existing applications available over the network. In early 1972, traffic over the network began to increase, as researchers and programmers alike began to make use of the network. By October, there were enough programs ready for use to capture interest of the crowds.

When the ICCC began, ARPANET was ready. The ARPANET group set up a section of tables containing dozens of computer terminals, upon which the thousand or so engineers in attendance could test the capabilities of the new network. Connecting to computer hundreds or thousands of kilometers away, even one in Paris, France, the users were able to try many programs, including meteorological models, and air traffic simulator, conferencing systems, a mathematics system, experimental databases, and a computerized chess game. The effect these demonstrations had on the engineers was extremely powerful. One attendee, Vinton Cerf, described the engineer’s reaction as, “just as excited as little kids, because all these neat things were going on.”17 It was apparent to all at the ICCC conference, that networks were clearly the wave of the future.18

As a result, over the next 20 years many changes happened to the ARPANET, until it was transformed into the Internet. In 1972, Ray Tomlinson of BBN created e-mail, and ARPA was renamed the Defense Advanced Research Projects Agency (DARPA). Development began on the protocol later to be called TCP/IP in 1973 by a group headed by Vinton Cerf from Stanford and Bob Kahn from DARPA. By 1975 it was complete. This new protocol was to allow diverse computer networks to interconnect and communicate with each other. The ARPANET continued to grow, with more and more nodes being added.

Simultaneously, other experimental networks began to spring up, including WIN, MINET, SATNET, AUTODIN I, and AUTODIN II. In 1982, ARPANET and the other networks switch over to the TCP/IP protocol, in order eventually allow greater connectivity. In 1982, the DoD split ARPANET into two seperate networks: a defense research network (still called ARPANET) and an operational military network, call MILNET.19

In 1983, the Internet was finally born, as ARPANET and some the surrounding smaller networks were joined through the TCP/IP protocol into one large network. While still technically under the control of the military, most military operations had been moved onto MILNET, leaving ARPANET purely the domain of academic researchers.

Between 1973 and 1989, the National Science Foundation (NSF) had sponsored and built it’s own small regional networks across the United States, between many of the Universities which were not connected to ARPANET. Near the end of the eighties, the NSF had created a TCP/IP backbone network called NSFNet between their small regional networks, creating a separate network which could compete with ARPANET. At the same time, The Internet was approaching the million-user mark, still using the original ARPANET backbone network. On Feb. 28, 1990, the original ARPANET was decommissioned, and the Internet was transferred to NFSNet. Upon the NFS aquiring the Internet, it finally became civilian-run.

Since then, the Internet has been privatized, and is now held by a large number of independent companies, all connected together through privately-held backbone networks. Use of the Internet has skyrocketed, especially due to the World Wide Web, developed on top of the Internet by Tim Berners-Lee in the early 1990’s.

Baran’s Influence in the Internet

Baran’s ideas mainly centered around the redundancy and survivability of a network in the case of a nuclear assault. By creating multiple routes and paths between points, if a path or two were to be destroyed, the network could still function. In the Internet, between any two points are thousands of paths. Due to it’s Baran-inspired redundancy, even if North America were destroyed, everywhere else in the world could still function. With the advent of communications satellites, even if an area were cut off from all ground links, it could still use a satellite to reach out into the surrounding world.

Baran also strongly advanced the notion of each switching station or node to be able to know how to route a packet, independently of information from other nodes using a store-and-forward design. In the Internet, every router knows where to send packets. Internet routers use a similar store-and-forward method to send packs, and use creative algorithms to decide the best place to send each packet next.

Many of Baran’s ideas have disappeared, however. Because he was working during the time of the Cold War, every aspect of his network was security- and military-oriented. This can be seen by his use of words such as “raid”, “salvos”, “target”, “attack level”, and “probability of kill” in his writings. He spent a lot of time writing about proposed encryption schemes for every packet to be transmitted, and burial of cables to prevent unauthorized access. The Internet, as we now see it is a major security hole. Natively, packets are sent unencrypted, and can be intercepted with very little effort. Some effort has been made to secure communications somewhat with the SSL and SSL2 protocols, but a completly secure solution may never be found, without a new implementation of the Internet.

Conclusions

Paul Baran had an indisputable influence on the development of the ARPANET and it’s modern descendant, the Internet. Many of the principles he wrote of and presented in the early 1960’s have lived on in the form of the Internet today. His early worked influenced the original creators of ARPANET; after 1967, he had direct “hand on” contact with ARPANET. Yet, to most of the world, he is unknown.

As one of the pioneers of network technology, he deserves a more hallowed place among the histories of Computer Science. He is not often written about in textbooks; He should be. His ideas provide a unique insight into the scientific mindset of his time, and a look at the constraints and military focus placed on scientific research during the Cold War.

1. The Daily, “Household Internet Use Survey”. (Statistics Canada: July 8, 2004), http://www.statcan.ca/Daily/English/040708/d040708a.htm (Accessed Dec. 7, 2005)

2. United Devices Cancer Research Project, http://www.grid.org/projects/cancer/ (Accessed Dec. 7, 2005)

3. Folding@Home Homepage, http://folding.stanford.edu (Accessed Dec. 7, 2005)

4. Andrew S. Tanenbaum, Computer Networks, 4ed. (Upper Saddle River: Prentice Hall PTR, 2003), 50

5. Paul N. Edwards, The Closed World: Computers and the Politics of Discourse in Cold War America (Cambridge: MIT Press, 1996). 133

6. Janet Abbate, Inventing the Internet (Cambridge: MIT Press, 1999). 10

7. (Abbate, 1999), 11

8. (Tannenbaum, 2003), 50

9. (Abbate, 1999), 17-18

10. Paul Baran, “On Distributed Communications,” (Santa Monica: RAND Report Series, 1964), Vol. V, Sec. II. http://www.rand.org/publications/RM/baran.list.html (Accessed Dec.8, 2005)

11. (Tanenbaum, 2003), 50

12. (Abbate, 1999), 36-37

13. (Abbate, 1999), 37-38

14. Davies et. al, “A Digital Communication Network for Computers Giving Rapid Response at Remote Terminals.” Proceedings of the First ACM Symposium on Operating Systems Principles. (New York: ACM Press, 1967) Pg. 2.1-2.17

15. (Abbate, 1999), 38

16. (Tannenbaum, 2003), 53

Bibliography

1.Andrew S. Tanenbaum, Computer Networks, 4ed. (Upper Saddle River: Prentice Hall PTR, 2003).

2.Paul N. Edwards, The Closed World: Computers and the Politics of Discourse in Cold War America (Cambridge: MIT Press, 1996).

3.Janet Abbate, Inventing the Internet (Cambridge: MIT Press, 1999).

4.Paul Baran, “On Distributed Communications,” (Santa Monica: RAND Report Series, 1964). Vol. V, Sec. II. http://www.rand.org/publications/RM/baran.list.html (Accessed Dec.8, 2005)

5.The Daily, “Household Internet Use Survey”. (Statistics Canada: July 8, 2004). http://www.statcan.ca/Daily/English/040708/d040708a.htm (Accessed Dec. 7, 2005) 6.Davies et. al, “A Digital Communication Network for Computers Giving Rapid Response at Remote Terminals.” Proceedings of the First ACM Symposium on Operating Systems Principles. (New York: ACM Press, 1967).

7. Vinton Cerf, Interview with Judy O’Neill, Reston, Virginia, 24 April 1990. Charles Babbage Institute http://americanhistory.si.edu/collections/comphist/vc1.html (Accessed Dec. 8, 2005)