Andrew Viterbi, an oral history conducted in 1999 by David Morton, IEEE History Center, Piscataway, NJ, USA.
https://ethw.org/Oral-History:Andrew_Viterbi
Viterbi received his BS and MS degrees from the Massachusetts Institute of Technology (1957) and his Ph.D. from the University of Southern California (1962). He worked on guidance systems for satellite spread spectrum technology at the Jet Propulsion Laboratory JPL (1957-62). He carried out theoretical and practical work on digital communications, publishing Principles of Coherent Communication (1966, on phase-locked loops and their application to analog and digital communications) and Principles of Digital Communication and Coding (1973) , especially the Viterbi algorithm for convolutional codes (completed in 1966, published in 1967). He taught at UCLA from 1963 until the late 1960s, then moved from part-time consulting to full-time work at Linkabit (which he co-founded with other professors). The company provides the military with digital, wireless, spread-spectrum communications technology with error correction and immunity to hostile jamming. Linkabit was acquired by M/A-COM in 1980, who sold its assets to other companies in 1985. Linkabit's technologies include VSAT, Video Scrambling (Videocipher) and Digital Satellite (Two-Way) Modems for the Armed Forces. Viterbi and Irwin Jacobs quit M/A-COM in 1985 and founded Qualcomm that same year. Qualcomm developed the OmniTRACS system, a two-way satellite communications system used by trucks for positioning and messaging, and its CDMA technology was later used for wireless communications. Qualcomm also developed the Eudora e-mail program for internal use and later marketed it. Viterbi concludes with a discussion of how good government support for basic R&D is, admiration for American universities, and criticism of American high schools.
Childhood and Educational Background
Morton: Tell me about your childhood. Did any influence lead you towards a career in engineering? I would also like to hear about your education.
Viterbi: I was born in 1935 and immigrated to the United States with my parents from fascist Italy in 1939. My family left Italy because the so-called racial laws were anti-semitic and against all races except the so-called Aryan race. We lived in New York City for a few years before settling in Boston when I was six years old. I basically went to every public school in Boston from second grade through high school. I went to a really good middle school called Boston Latin School. It is the oldest school in the United States, dating back to 1635.
Morton: Did you know English then?
Viterbi: I learned English when I was four years old and it was a bit difficult in kindergarten. When I was in first grade in NYC, I was already pretty fluent. I never had any difficulty in Boston. I don't believe that English is a second language, but I don't think we need to discuss this topic in depth. Going to MIT has been a dream of mine since I was ten years old, looking across the Charles River. I studied at MIT for five years and got my master's degree. It's now a standard course for electrical engineering and computer science majors at MIT, but at the time it was reserved for corporate co-op students. The Massachusetts Institute of Technology has a training plan to obtain a bachelor's and master's degree within five years, combined with one or one and a half years of corporate work experience. In order to balance work and school, classes can be held in the evening. As refugees, we were financially limited at the time, and as far as I was concerned, this was the ideal course. My father was a doctor, but had very limited opportunities to practice.
Morton: Where do you do your corporate work?
Viterbi: At Raytheon.
Morton: Did you work on a specific project?
Viterbi: It's a great development program. I originally worked in the semiconductor field. In those days it was brand new and was called the transistor division. That was 1954. I was involved with what they called communication equipment, which was radio links and some early television and CCTV. I actually designed some circuits. Then I went to a joint venture with Honeywell in Minneapolis called Datamatics. They competed with RCA to make some very early computers. Believe it or not, it was still tube based back then. I'm not into hardware, but really interested in software. I wrote some diagnostics and such. This is a fairly broad set of tasks. I can't say it had a huge impact on my later industrial experiences, but it opened my eyes to what engineering is all about.
Jet Propulsion Laboratory
Morton: You went to the Jet Propulsion Laboratory (JPL) in 1957.
Viterbi: Immediately after my master's degree in 1957, I went to JPL in June of that year. That was three months before the launch of Sputnik (the first Earth satellite developed and launched by the former Soviet Union). I had the privilege of being a part of the first successful U.S. satellite program, Explorer 1. More importantly for my subsequent experience and career, I have been exposed to spread spectrum technology from day one of my full-time engineering job. In 1957, neither NASA nor the space program existed. JPL is owned by Caltech, but is operated under contract with and fully funded by the U.S. Army Ballistic Missile Command. It was this command that financed the Redstone Armory under Werner von Braun. We did the communications, command and control for the missiles that Redstone was building. We work under Caltech professor Bill Pickering. That's a radio inertial guidance system using spread spectrum technology. That was the first time I learned about things like shift register sequences, which have been very useful throughout my career. When the Russians launched Sputnik in October 1957, JPL almost immediately began turning to the space program. The US space program was handed over to NASA sometime in 1958, but Explorer 1 launched in March 1958 with the integrated package. These communications are primarily carried out at JPL. We call it telemetry because it's actually measuring distance. While I did some spread spectrum work at JPL, I worked mostly with tracking and receiving equipment, specifically a type of equipment called a phase-locked loop. It was brand new then. It's been used commercially in color television, but more for coherent communications tracking that JPL is doing. This work is more advanced than that in many other places, including MIT's Lincoln Laboratory.
PLL technology and analog and digital communication
Morton: PLL technology is essentially for analog communications, isn't it?
Viterbi: Digital communications also need it. The receiver must first be synchronized to the carrier and must track that carrier. Second, it must be synchronized with the bit timing. This is called a delay locked loop, but it's all the same technique.
Morton: Did they use digital communications from the beginning?
Viterbi: That's a good question. A lot of it is simulated, there's no question about that. There may not have been any digital technology in Explorer 1's communications. Communication is not used as much as tracking and commands. Orders are of course digital. The digital transfer of space vehicle communications may not happen until later. By 1960, we were doing a lot of theoretical research on digital communication and doing the first experimental work. What I originally did at JPL, and which JPL was very deeply involved in, was tracking and synchronization. My first book, published in 1966, was Principles of Coherent Communication. Half the book is about phase-locked loops, and the other half is about its application to analog and digital communications. By 1960, I was very interested in digital communication, transmission, modulation, error correction coding, error probability analysis, etc. I wrote the first papers on digital communications in 1960-62, which, in general, were widely recognized. I don't have a PhD yet, so I'm taking some courses, but mostly I want to write a PhD thesis at USC. They have some really good and very dedicated teachers, especially in mechatronics and controls. Communication is a bit weak. I also have excellent mentors at JPL who are currently professors at USC.
Postgraduate Study and JPL Jobs
Morton: You graduated from USC in 1962. Did you start working at JPL before entering graduate school?
Viterbi: No, I started working at JPL while going to graduate school. My first choice for grad school was Caltech, but Caltech didn't want me unless I went there full-time and didn't work at all. I was working at Caltech, but Caltech was very snobby about JPL at the time. People at JPL are considered workers. They won't allow one person to do both at the same time.
Morton: Why did you move to the West Coast after graduating from MIT?
Viterbi: I loved Boston and still do, but I'm tired of it. I still go there a lot, I have a lot of connections with MIT and even my high school. My last conference call was a fundraiser for the Boston Latin School. My family is refugees, we are special refugees because we are Italian Jews. There are many Italians and Jews, but Italian Jews are almost non-existent. We made some friends, but no one very close. Driven and attractive. The driving force is the weather. The weather in Boston is terrible. My parents are getting older and I think the west coast weather would be better for them. We also have relatives in LA, my mother's sister and some cousins, so that was another draw. The West Coast's openness to industry may be just as attractive compared to the East Coast. Many of these problems have been alleviated over the past forty years. Raytheon was a very good company, as were GTE Sylvania and many other communications companies, but the opportunity to work for what was then Hughes Aircraft or TRW or a more academic institution like JPL seemed more exciting and the working conditions were better . East coast industry put many engineers in the bullpen. The same goes for the aerospace industry, including General Dynamics and Lockheed.
Simon Ramo and his colleague Dean Wooldridge are founders of TRW, and Ramo wrote The Business of Science. They previously served as vice presidents of the Hughes Aircraft Company. They're really managing Hughes because Howard Hughes is busy making movies. They greatly improved the quality of life and professional status of engineers on the West Coast by giving everyone an office, or at least a limited number of people in each office. At JPL, we have four people in one office. The people I share the office with are really nice and a pleasure to share the office with. Also, it's not a big bullpen. In my opinion, because of these, the east coast lagged behind the west coast and didn't really catch up until the mid 70's. Companies like HP eventually spawned an entrepreneurial spirit throughout the Bay Area, but that doesn't exist on the East Coast. Regardless, it started on the West Coast and was not emulated in places like Boston, North Carolina, and New York until much later.
Morton: I got off topic. Let's go back to JPL again.
Viterbi: I completed my Ph.D. in Digital Communications at USC. My thesis wasn't my best work, but it was solid and got me thinking. Once I get the paper out, I start doing good work. About a year after I got my PhD. I had the opportunity to join the UCLA faculty. This is a great growth opportunity. For many years after that, I continued to advise JPL one day a week. I learned a lot at UCLA because I had to teach. Early on, I had a considerable interest in information theory. Although I've been doing digital communication work and theory, I really don't know enough about information theory. Shannon's theory was first developed at Bell Laboratories, then MIT became the center. Bell Labs is still going strong, but I think MIT became the center. Unfortunately, I didn't take any information theory courses at MIT because I left too early.
Viterbi Algorithm
Viterbi: I started looking at the information theory work that had been done and started teaching it. That was a great learning experience, when I developed what is called the Viterbi algorithm for convolutional codes. It comes from my teaching. I have something very difficult to teach, very difficult to teach. When doing research, people get into a mindset where they think intuitively, but when it comes to teaching others, it becomes clear whether they fully understand it. I found information theory difficult to teach, so I started developing some tools. This algorithm came out of there.
I wrote my first paper in March 1966, but it was not published until April 1967. It demonstrates some properties of convolutional codes. I didn't provide algorithmic details at first. Fortunately, one of the reviewers, Jim Massey, encouraged me to join the algorithm. (Jim Massey was a good friend of mine and one of the major contributors to information theory, coding, and cryptography. He was at Notre Dame at the time.) It's a clumsy description, but at least it establishes itself. No one thought it had any potential practical value, because my estimate at the time was that it would take a few thousand registers to make it work. About a year later, my colleague Jerry Heller at JPL did some simulations and found that it could be done pretty well with just 32 or 64 registers. This makes it more practical. However, it's still a big monster, filling a rack of roughly 18-inch cases. Now it's just a small part of the chip.
Morton: Where did you originally publish this article?
Viterbi: In IEEE Transactions on Information Theory. The second paper I wrote was also published in the April 1967 issue of Communication Society Transactions. I think it was called IEEE Transactions on Communication Technology at the time.
Signal Processing, 1960s and 1970s
Morton: Were you connected to the emerging digital signal processing community at the time?
Viterbi: There really was no digital signal processing community in 1967. There were no microprocessors back then. Sure, some people considered RISC machines in the late 60's, but everything was built from components. There is some level of Small Scale Integration (SSI). It's a term that no longer exists. In those days we talked about SSI, MSI (medium scale integration) and LSI (large scale integration). At that time, very few people did LSI. What we call SSI today may have been called MSI back then. I'm getting a little ahead of myself, but the first chip designed and built by our first company, Linkabit, was a hundred transistor integrated circuit. That was in 1973 or 1974. It is actually a signal processor for the Viterbi algorithm. Linkabit almost went out of business as several fabs went down. In 1967, there were no such signal processing chips. There were devices that implemented arithmetic units, and that's what our chips later became. If I recall correctly, Intel introduced its first microprocessor, the Intel 404 or 4004, in the early 70's.
Morton: I think it was around 1975.
Viterbi: That's not a signal processor, it's an arithmetic processor. If you look at IEEE history, what is now the Signal Processing Society was then called the Audio and Acoustics Society.
Morton: It has changed names several times.
Viterbi: I remain interested in things like control theory and signal processing because it's so closely related to communications.
Industrial, Government Contracts
Morton: Based on the fact that you liked JPL and then went into teaching, it seems from the first part of your career you were drawn to the theoretical side of engineering, but then you started Linkabit in '68 or '70.
Viterbi: I'm not completely out of industrial applications. While at UCLA, I continued to consult at JPL one day a week and maintained a very close relationship with the space program. Even the algorithm, when I think about where it's applied, it's obviously a broadband Gaussian channel in my mind, the spatial channel. This is where most of the early work was done. JPL embraced it quickly, much faster than any other research lab or company. I also consult for quite a few companies. I can barely remember going back to the 60's, but ITT Defense Communications and Ford Aerospace were some of the companies I consulted with. I also consult with a number of small companies in the Los Angeles area. Regardless, I keep getting opportunities to see the system in action.
During that time, I wrote my second book, Principles of Digital Communication and Coding. I actually wrote most of it while I was at UCLA, but finished it around 1973. My colleague Jim Omura, who was still on the faculty at UCLA at the time, and I wrote the book together, about a quarter of the way through. This application problem has always existed, and opportunities abound to apply it to actual defense and space communications systems at NASA and the government.
In the late 60s, Irwin Jacobs at UC San Diego, my colleague Len Kleinrock at UCLA, and I decided to pool our consultations. There were a lot of opportunities at the time, including opportunities to get small government contracts. Our first contract was with the Naval Electronics Laboratory in San Diego. We study error-correcting coding and its application to naval communications. Later we did similar work for NASA. None of us could run the company alone, and we saw that we might need a full-time person, so we formed Linkabit in 1968. In 1969, we hired Jerry Heller as our first full-time employee. It's not like jumping into the sea suddenly, but more like slowly getting our feet wet. In '68 or '69, very little happened except writing some reports. We probably built some hardware in 1969 and developed software for the Navy computers. By 1970, we were building prototypes for NASA.
Morton: Is the Navy still using it for space communications?
Viterbi: No, the first job in the Navy was actually working for HF. It is ground or marine high-frequency digital communication with a message speed of 1200 bits per second.
Growth in digital communications, satellite communications
Morton: I think digital communication is proliferating.
Viterbi: Yes, it started proliferating in the 70s. There are two things going on. First came cable development, where modems were still in their infancy. If you have a PC at home with a built-in modem that connects to a phone line, we call it an analog modem, but it's actually a digital signal generated on an analog line. This already happened in the late 60's. In 1970, the Communications Society held debates and famous symposiums, concluding that encoding would never have any application in commercial systems, and even data modems would never get very far.
Morton: I want to find it somewhere.
Viterbi: Bob Lucky, after hearing a couple of speakers say that coding is dead, mostly in jest, stood up and said "well, data is dead too." It's the same theme. On the wireless side, probably only NASA and the military are really doing digital communications. The military needed to have a signal that was immune or resistant to hostile interference, so they chose spread spectrum, and using spread spectrum required digitization. Setting up an analog signal on spread spectrum doesn't make much sense. People have seen it, but it's never really been practical. However, numbers are natural because spread spectrum waveforms are digital. The simplest form of spread spectrum is to beat the center frequency very rapidly over a wide spectrum. The military was doing this even in the 50's and even more so in the 60's.
It's satellite communications that's really driving everything, whether it's for the military or for the space program. Now, we are improving the efficiency of transmission from great distances. These are geosynchronous ranges, which means 40,000 km. All the technology used for military spread spectrum becomes even more important because a geosynchronous satellite is a sitting duck and anyone can jam it, whereas on the ground antennas can usually be positioned to avoid jamming. Commercial satellites were originally analog and by the 1970s were using digital technology. Digital communications got a big boost from satellite communications. Some of the legacy from digital satellite systems in the 60s and 70s to some cellular systems in the 90s can be traced back to this. I'm not talking about spread spectrum. I'm talking about something like TDMA, which was done on satellite in the first place for good reason. They are then imported into the hive for much less reason.
Linkabit and M/A-COM; Development and Consumer
Morton: Is it tough to compete with a company like AT&T that has its own in-house R&D and a lot of experience?
Viterbi: Linkabit didn't compete with AT&T, basically since 1970, Linkabit was a very small military contractor. In 69 we had a small office in Los Angeles, and in 1970 we moved to the Sorrento Valley and we had our first 4000 square feet. There are ten of us, pursuing few government contracts. AT&T and Bell Labs didn't pay attention to us. Because of our professional reputation, we meet with many people at Bell Labs. I've been to Bell Labs a few times and given some lectures there, but that was more academic than competitive technology.
There was no such thing as intellectual property back then. It's just amazing. you do not need to worry. What happens to the algorithm is interesting. We got a lawyer to register the company, and shortly thereafter we got some work on algorithms from NASA and the Navy. We went to the same attorney, who is also a patent attorney, and asked, "Do you think we should patent this?" His response was, "This is ridiculous. These things are only for government applications, do you really There's no protection. The government can do whatever it wants." That's an overstatement and not entirely true. He convinced us that it wasn't worth spending three thousand dollars or anything to prepare and file a patent.
Morton: Do you regret it now?
Viterbi: No. If we had patented it, it might have slowed its acceptance because no one was patenting it at the time. AT&T and IBM patented it for commercial reasons, but we are a small government contractor. Throughout the '70s, we started thinking about commercial applications, but it didn't really go commercial until 1980 when Linkabit was acquired by M/A-COM, a small conglomerate headquartered in Burlington, Massachusetts. Then, we did a lot of things with great commercial success. One of these early successes was VSAT. We took what was essentially a military technology and applied it to commercial satellites. Satellites are born. This is where the transition is easiest.
We have a contract with American Bell Iran International (ABII), which is actually a subsidiary of AT&T. In the mid-70s, AT&T awarded the Shah a contract to redesign and renovate Iran's entire telecommunications infrastructure, much of it via satellite. There is a thing called IranSAT, we built a transmit and receive modem, and the Iran SAT satellite communicates at high speeds. In early '79, I went to Bell Labs in Holmdel, NJ and asked if there was a problem. The program manager told me, "I think the program will survive. We don't worry about politics. The revolution was in March, and they wake up around May or June and say, 'Maybe we should cancel this program.'" What is the total cost you will have to be reimbursed if we cancel? Our finance guy did an honest assessment and it came out to be around 90% of the total cost. They said, "In that case, you might as well go ahead and finish it. We're going to use it somehow. I don't remember how, but we delivered it. It turns out, it's a sort of prototype of VSAT, and we're at Serious research began in the early 1980s.
By this time, we had the funds to invest in commercial development. M/A-COM is a much larger company with less than $1 billion in revenue, while Linkabit probably had $25 million in revenue at the time of the acquisition. Even more interesting, we developed the first video scrambling system, a digital video called Videocipher. This continues to this day. It was originally designed for Home Box Office (HBO), of which he was a client. By the early 80's they started sending analog transmissions to all cable heads. They call it the primary video service. When the service started, the only organizations with three- or four-meter antennas were the cable companies, and it was assumed that they would be the only ones. By the early '80s, however, wealthy ranchers in Texas and elsewhere started getting the shows. Then, when the price dropped from $100,000 to $10,000, people in rural communities started buying them until there were about half a million three- and four-meter backyard dishes. They got HBO's signal for free, so HBO was pissed and decided to encrypt their signal.
We at M/A-COM Linkabit have developed what we consider to be a very secure system, but it requires full digitization of video and audio. We could have shipped two or three thousand of these for $3,000 to $5,000, but Backyard Dish owners have a powerful lobby group called SPACE, Private Association, that lobbyes Congress. Their argument is, "It's not fair. We're willing to pay for the service, but they can't take it away from us. They have to give us a descrambler. We'll pay for the descrambler, and we'll pay for the service The provider pays something. I'm not sure if they actually say that, but that's the way it is.
The problem is that descramblers can't be sold for $3,000 a piece to people who might only spend $5,000 on groceries. The device needs to cost between $200 and $400. So we had to compromise and completely redesign from a more expensive, professional device to a cheaper consumer device. We can still digitize and hard scramble using Digital Encryption Standard (DES) on audio, but on video, we have to do something else. Digitizing video means using very fast A-to-D converters, and those cost too much. So we simply rotate each line so the lines don't line up. This keeps costs down so that descramblers can be sold for less than $400. A key step in this, by the way, is building an application-specific integrated circuit (ASIC), which is a signal processor. This is our first major ASIC development. We've done some mid-scale integration for our decoder work, but I don't recall us doing anything at the scale of that device. That was in the early 80s.
Morton: You bring up the importance of the consumer in designing a product that was originally intended to be irrelevant in the consumer market. In this case, the consumer market has a big influence.
Viterbi: The same is true for VSAT. They are not professional equipment, but they were originally intended for industrial use without consumer needs in mind. We did other things like networking multiple terminals sharing a mainframe. We did this in the early 70's just for internal use. By then, the company had about a thousand people, and we had close to a thousand VT-100 terminals. Instead of PCs, they were dumb terminals that connected to VAXs made by Digital Equipment Corporation (DEC). As a company, we primarily sell to digital equipment companies and ultimately sell networking products to DEC.
Around the same time, things started to fall apart at M/A-COM. M/A-COM doesn't really exist anymore, just as a division of AMP. The original chairman and CEO of M/A-COM, which acquired Linkabit, was forced to leave, and the new leadership lacked vision. So some of us left in the spring of 1985. Interestingly, M/A-COM subsequently sold all of its businesses. They first sold the VSAT business to Hughes for $100 million. This turned into a half-billion-a-year business within a year or two. They also sold some very early wireless mobile phone work that we did. It doesn't have to be mobile, but it's wireless. They sold all of them to Hughes. Hughes bought all of Linkabit's commercial satellite business as well as Digital Communication Corporation, another division of M/A-COM in Maryland. From these two divisions Hughes started Hughes Network Systems and it was very successful.
Another piece M/A-COM sold was the video scrambling business, which was acquired by General Instruments (GI) and a cable TV plant for about $250 million. This generates billions of dollars in annual revenue for GI. M/A-COM later sold the government production as well. We're doing quite a bit of government work. At that point, maybe a third to a half of the work we have left is done for the government.
Dual Modem and Spread Spectrum Modem
Witte: While we made a lot of error-correcting equipment for the government, the most important government product was the first microprocessor-implemented digital satellite modem, called a dual modem. Irwin Jacobs is more lyrical than I am at this. Dual modems have been manufactured for the USAF since the mid 70's. By the end of the 70s it became a production program and continued into the 80s and 90s. It is still being built by a company that later bought the Linkabet government unit. Initially, the dual modems were used in the Air Force SATCOM program. This is something the Air Force can use from the command post to communicate with the bomber fleet, B-52, FB-111, etc. Among other things, it sent an emergency action message that was never sent, thank God. However, that's what it's for. Rockwell originally manufactured this product in the 60's and early 70's. Although it is a digital transmission, it is largely an analog implementation, essentially an FM modem.
The Air Force realized they needed to modernize, especially to introduce more security, so spread spectrum came into the picture. We built a spread spectrum modem. It has to be dual, that is, it has to be backwards compatible with previous modes, and it has to be smart. Instead of implementing a microprocessor on a single chip, we implement a microprocessor on a standard commodity chip. There was already highly integrated memory and some early signal processors, and we built a little computer out of these types of components. It's actually a RISC machine that implements all the algorithms to do spreading, modulation, encoding, decoding, etc.
It took us about three years to convince the Air Force that (a) this was doable, and (b) this little company called Linkabit could do it. We ended up building thousands of these dual modems. After the Air Force, we designed them for the Navy and Army, each with their own capabilities or requirements. Army modems became triple modems instead of dual modems. There's a company in San Diego called Titan that bought M/A-COM's government systems division that used to be Linkabit's and they're still building this stuff. To a large extent, this work made Linkabit, probably more so than the work done by coding. Linkabit is a research and development company and manufactures in limited quantities. Once Linkabit was acquired by M/A-COM, we became even bigger.
Qualcomm and UC San Diego
Viterbi: In the spring of 1985, Irwin Jacobs and I quit within a week. After relaxing for two or three months, we founded Qualcomm in July 1985. Many immediately joined us. We don't know what we'll do. I have been teaching. After moving to San Diego in '73 and becoming a full-time employee of Linkabit, I officially resigned from UCLA in '75, but I stayed at UCSD part-time. In '85, between companies, UCSD gave me a part-time fixed appointment because I felt I wanted to do other things too. For the next ten years or so, I was a quarter professor. In total I have been in the UC system for about thirty-one years, although only ten of those were full-time. I've been kind of busy doing this, but when the opportunity presented itself, the people I used to work at Linkabit, who were actually elite, were too motivated to refuse to join. Irwin Jacobs would say we start in his study, but I'm not actually there because I'm traveling in Europe on a really nice cruise ship. I'm back in mid-July. Back then, we had a palatial office in a cleaning shop that was at least 3,000 or 4,000 square feet. We've contracted for a building on Sorrento Valley Road, but it's not ready yet. We still own that building. Qualcomm started much faster than Linkabit, and we started in a dentist's office on the edge of UCLA.
I'll tell you the Qualcomm story shortly. It started with about seven people and a secretary. We did not consider commercial work on the first day. We pursued a mid-sized government job and had a fairly large data link contract. It's a joint service program, Army/Navy/Air Force, for a test range. They need very fast telemetry and messaging to and from aircraft, tanks and other mobile platforms. Then we did some satellite projects. One satellite program was for what was then Ford Aerospace and now Loral, and the other was for Hughes. By the late 1980s, the Department of Defense budget was depleted. We did a very interesting project in low earth orbit with multiple satellites. Globalstar and Iridium are descendants of this concept. But it's for the military. The program was canceled due to budgetary reasons and the aerospace company's dislike of the concept. They only face geosynchronous satellites.
Before Qualcomm was officially formed, I got a call from a colleague who worked at a company called OmniNet, which was trying to provide satellite location and messaging for the transportation industry. It's a long story, but in the end OmniTRACS came out of it. OmniTRACS is a two-way satellite communication system.
Morton: Is this what truck drivers use?
Viterbi: Yes. There are currently 300,000 trucks, of which about 260,000 are in North America, and the rest are especially distributed in Europe, but also in Mexico, Brazil, Korea, Japan, Malaysia and now China. About 25,000 or 30,000 trucks in Europe use the system. It's been a long process, and I'm going to cut it short. It took about a year to develop a prototype, but the prototype is just one way, from the satellite down, or should I say from the hub to the satellite to the truck driver. It didn't go the other way because there was another company called Geo Star that made it go the other way. Our client, OmniNet, tried to work with Geo Star, but couldn't.
Then we went in both directions. Interesting that everyone else is trying to use L-band and process satellites, but we're using Kl.c. bands, the 12 and 14 GHz bands. There are very few L bands. If you want a processed satellite you have to have your own package, while there is a huge supply using non-processed satellites with the Kl.c. band. Beyond that, some direct satellite radio services were introduced in the 80s but didn't take off. Therefore, our customers are able to rent direct satellites at a fairly cheap price.
Developing this product has been a remarkable achievement. It involves a lot of signal processing and a very clever antenna with directional antenna control that moves with the truck and always faces the satellite. It's mostly Irwin Jacobs' conceptual approach, and some really good development work by others at Qualcomm. It all came together in early '88, but that's when our client went bankrupt. Our only option is to buy them for inventory. It was a very expensive proposition, but it worked out well because the business did well in the long run.
We have a great customer, Don Schneider of Schneider National. At the time he had about 10,000 trucks. Schneider probably has 15,000 to 20,000 trucks today. Our business was built from there. Even after Schneider, we had a drop because there wasn't another Schneider around. We approached JB Hunter, but Hunter was not that receptive and waited a few years before joining. OmniTRACS is finally profitable. We certainly didn't recoup our investment, but our revenue now exceeds the cost of operating and building the system. Today, OmniTRACS is used by at least 90% of companies with more than 1,000 trucks, and it is a very successful enterprise.
Then one of our very smart people, Klein Gilhousen, came in and said, "We're using CDMA. We're using spread spectrum for the system. I forgot to mention the most important thing, that's why we're using CDMA, why we're able to Use where others can't use these low cost satellites. Low cost satellites aren't really meant for mobile use. The FCC license for the Ku band is only for fixed satellites and fixed transmitters/receivers, and for mobility, L-band should be used.
However, there is a footnote in the regulations saying secondary users can be mobile. Secondary users can use satellites, but only if there is no interference. The distance between satellites tends to be about 2 radians, and the mobile terminals are very small, so the antenna aperture is about 6 to 12 degrees, so they splash signals onto other satellites, so naturally they interfere. Instead, they are left wide open so that other satellites can interfere with these terminals, and they have no recourse. Either way there is no recourse. Too bad if other people interfere with secondary users. If secondary users interfere with others, they go off the air.
We have to convince the FCC that we can build a system that doesn't interfere, and that's a huge struggle. However, the FCC is very liberal with experimental licenses, they gave us licenses for about 600 trucks, not experimental ones. It took at least a year and a half to convince the FCC that we didn't interfere. Then they gave us licenses for 20,000 trucks. Today, we have about 300,000 licenses in the U.S. and will grow to about half a million, and we've never had a case of interference. Our mobile transmitters send a watt across transponders about 24 megahertz wide, so our signal looks like noise. We are heavy on the noise.
Another aspect related to our competition is that when a satellite is leased, it's leased from a provider like GE Satellite. They coordinate. Unless the next satellite is transmitting full blast TV, we can live with it if it's just a VSAT. It looks like interference, but it's spread spectrum because in the process of despreading the desired signal back to the original narrowband, another signal is spread and looks like noise. Regardless, we make commercial spread spectrum a reality by loading tens or even hundreds of thousands of users.
Applying Spread Spectrum to Cellular Communications
Viterbi: Then Klein Gillhausen came in one day and said, "Why can't we use it for cellular? There's a lot of interference out there. That makes sense to me. I gave the IEEE Communications Symposium in '82 Submitted a conference paper saying "this spread spectrum might work for cellular" but no one was listening, that's not what M/A-COM Linkabit is about. When Klein suggested this, I said:" This sounds familiar, but you have a power control problem. You have to make sure no one is drowning your signal.
The first thing Klein, Owen, Butch Weaver and I did was power control. We've come up with a really clean and neat approach that combines what we call open-loop and closed-loop power control, and it works really well. At first, no one believed us except Pac Tel Cellular, now Vodaphone Air Touch. Pac Tel listens and lets Bell Atlantic, NYNEX and Ameritech listen. They all made a million dollar investment in us. That's a small amount now, but it was a lot of money back then. They helped us pay for development and we were up and running. I'm not going to give you the whole CDMA story. I can give you three papers on this subject.
Eudora Email Product Development
Morton: That would be great. Transitioning from satellite to cell phone and tracking made sense to me. I also saw Qualcomm's name on an email product called Eudora.
Viterbi: Yes. It's an interesting coincidence. As I mentioned, both at the old company and at Qualcomm, before the Internet or intranets exploded, we were all intranet oriented, we built our own email product. We found out that there was a guy named Steve Dorner in a supercomputer lab at the University of Illinois who wrote the program, and we licensed it. In the end we hired this guy. He still lives in Champaign Urbana and works for us full time. A guy named John Noerenberg was an email software engineer at Qualcomm, and I think it was his idea to market it.
We commercialize it a little at a time. Initially, we sold it for $25, which was a nominal amount. We have two versions. One of those versions is Eudora Light, which we're giving away online. Another version is the Eudora Pro that we sell. Before you know it, the combination of the two has 10 million users. Every government lab and every university seems to use it. Even in Europe, I had people ask Qualcomm if it was the one that made Eudora, and I said, "Yes. It's a side business. Our main business is cell phones and satellite systems.
Morton: Does Qualcomm make money from it, or do most people use the free version?
Viterbi: It's huge, relatively cheap publicity. Initially we made a little money. We started marketing it for real and lost money. Then it goes through a kind of quiescent period. Eudora 4.2 is really pretty good. Now you can easily find all the old messages you discarded and things like that. We are now supporting products with real e-commerce ads. People can have Eudora for free, as long as they accept the advertising message. This is in alpha testing and will be beta testing soon. Technically, this makes sense.
Whether it makes sense as a business is uncertain. It could really take off, but if it doesn't take off, that's okay because that's not our main business.
federal government research and development funding
Morton: The federal government has benefited your various companies in different ways over the years.
Whitby: Exactly.
Morton: How do you feel about government spending on R&D and procurement?
Viterbi: I'm a member of President Clinton's Advisory Council on Information Technology, and six months ago we just released a report urging the administration to continue basic research. Not application-oriented research, but basic research. No one has yet developed the transistor of the twenty-first century. The research that led to ARPAnet and Shannon's information theory fueled our information economy, the fastest growing segment of our economy. Basic research also gave rise to transistors and radio astronomy. This kind of research would not be conducted by the business because the shareholders would not allow it. GE and RCA abandoned pure research three decades ago. Bell Labs and IBM gave up on pure research about five to ten years ago. None of them do real basic research. If yes, that's also trivial.
As for where it should be done, I think it should probably be done in college. They are ready for this. CEOs can't turn to shareholders and say, "I'm creating shareholder value by doing research that may or may not have an impact a decade from now, and I may not commercialize it." That's where IBM is, and of course AT&T Found myself in this position. In the past, AT&T had the ability to have Bell Labs do pure research because they were a monopoly. There was a lot of forward-looking research out there, with no constraints. AT&T could guarantee a 7% return to shareholders, if it wasn't for Bell Labs, maybe 7.5%, but they're in no position to complain.
Morton: How do you overcome the objection that, as the head of a large corporation, the federal government is funding research that threatens your own? What if they tried to invent the next generation of wireless communications?
Viterbi: If you can't take advantage of fundamental research quickly, you don't belong here. Then you have no right to be a most important and successful company. If you're talking about wagon builders going out of business because someone figured out the internal combustion engine, that's a cliché. No one has the right to say that today.
We're doing some very innovative things, like a system for high-speed wireless access to the Internet, and we'll be announcing that in a week and a half. We have speeds of up to 2 megabits per second, the wireless equivalent of DSL and cable modems. We use cellular frequencies with a bandwidth of about 1.5 megahertz, with a transfer rate of 2 megabits per second. It's been quite a challenge, and I think we've demonstrated capabilities. This is very well developed in advance. It's also good applied research. However, it is not basic, fundamental research. We hope this will be a major commercial success within two years.
Basic research is purely speculative and can often take five to ten years before any application is developed. Five years is optimistic. More realistic ten years from now. We should do it in our national interest. After all, what do we have that the rest of the world doesn't? We have a very good higher education system and a rather weak secondary education system. It's a mystery how we can have the best university in the world that everyone wants to come to and have a pretty weak K-12 program. The answer is ridiculous. Because all the rest of the world wants to come and study here, we admit a certain percentage, maybe less than 10% of applicants, and they maintain the quality of our higher education. This is what I said at my graduation ceremony at Berkeley.
Morton: This is asking for trouble.
Viterbi: A faculty member told me that this is a very controversial subject. Part of it is I said that if someone graduates from Berkeley Engineering, they are good, so all the parents of the students are excited. Having said that, I would say that what has brought us a higher standard of living is the fact that we can take technology that has been built over the years and apply it quickly. This is because of venture capitalist funding, because we are much faster than the Europeans, because we are more entrepreneurial than the Japanese. The Japanese are fast, but monolithic. They often go in the wrong direction, like the Muse HDTV system, an analog system that was made obsolete by digital technology. Digital cinema is another system development here. This is all great for the next decade or so, but what happens next?
Morton: Right.
Viterbi: Everyone else is catching up. Europe seems to be doing this in an odd way. They tend to close themselves off by having their own standards. They try to do what the Japanese did 20 or 30 years ago. The Japanese don't do that anymore. China is another matter, but I've gone too far.