I wrote last
week about the results of the three sabbatical leaves I took during my
academic days. An industrial leave
had an even more profound impact.
In 1968, a stranger came into my office at UC Berkeley. He said that he and a partner were
starting a company which had the possibility of a contract from the U.S.
Department of Housing and Urban Development. He had been referred to me by one of my academic colleagues
as someone with technical knowledge that might help in the contract. That was true, and I began consulting
for the new organization, which was called Teknekron.
HUD wanted to explore technologies with which the locations
of vehicles in large fleets—buses, police cars and fire engines, taxis,
delivery trucks—could be monitored in order to improve the efficiency of
dispatching them. (This was well
before the Global Positioning System of satellites was conceived.) One method could be radio
multilateration, i.e., measuring location by comparing the times of arrival at
dispersed listening posts of a radio signal from the vehicle. Radio-signal propagation in urban areas
was encompassed in theoretical research I had been doing for more than a decade.
I proposed that, instead of actually constructing several
particular multilateration systems and evaluating their relative performances,
Teknekron should focus on measuring and modeling the characteristics of urban
radio propagation itself. Once we
had a good model, it could be simulated on a computer, and various
multilateration systems could be compared in simulo to determine which might be most successful. I contended that computer simulation
would be much less expensive and more comprehensive than physical trials of a
number of "in the metal" systems.
That approach was adopted, and I began designing an
experiment that would lead to the urban-radio model. It soon became clear that I was needed full-time to lead the
experiment, so I took an industrial leave from Cal in 1969-70. The experiment consisted of circulating
through typical types of urban and suburban areas in an instrumented truck, and
measuring, at a prescribed set of locations, the distortion of radio pulses
transmitted from an elevated position.
Here's a much-younger me in the truck recording some of the data:
 |
| Taking data. |
And here's a sample of what appeared on the oscilloscope in front of me:
 |
| The responses of the urban radio medium to narrow pulses at three frequencies (logarithmic scale). |
(Amusing aside: The project went under a rubric that encapsulated its purpose, "Public Urban Locator Service" or PULSE, which was emblazoned on the side of the truck. In the Berkeley of the late 1960s, we should have known better. The constant appearance in the streets of a truck so-labeled, bristling with antennas, stoked an inevitable Berkeley paranoia: were we part of some nefarious project that was tracking innocent people? We attracted much attention and some protest.)
The waveforms from thousands of oscillograms like that above
were laboriously manually digitized, yielding some 100,000 punched cards of
data—the standard storage technology of the day. The data were then analyzed and transformed into a statistical
model of the urban radio medium.
When experiments using the model were run in simulo and the results compared to those of equivalent
physical experiments reported in the literature, we were ecstatic at the superb
agreement. We had indeed
successfully captured the urban radio medium in a computer! I breathed a deep sigh of relief.
Thus having eliminated the need to build actual
multilateration systems for physical trials, we moved to the main task of the
project: running simulated trials of such systems and calculating the location
accuracies that could be expected with each. The results gave us insight beyond all our
expectations. All told, it was
perhaps the hardest year of work I had ever experienced, but well justified by
the outcome.
Teknekron went on to try to commercialize the most promising
of the systems we had simulated, but was decades ahead of itself in terms of
viable and inexpensive electronics.
Commercially practical multilateration of the type we simulated in 1970
had to await the late 1990s. It is
now a commonplace, employed by the cellphone industry to locate cellphones with
excellent accuracy. And the
totally different GPS technology is of course also available; its users
anywhere in the world can locate themselves with it.
I went back to Cal much enriched professionally by my
industrial leave. The 100,000
punched cards were a trove of still-unexamined data, which led to several PhD
theses under my supervision and a further-enhanced model of the urban radio
medium. That model was used in the
design of the GSM cellular telephone system, now one of two worldwide
standards.
On comparing the sabbatical leaves I described last week
with my industrial leave, there is no contest. While the sabbatical leaves might have been more relaxing,
even more creative in terms of theoretical work, the practical work of the year
at Teknekron was a new and immensely valuable departure for me, one that I
couldn't have done at the university.
For engineers and scientists, I think that a revolving door between the
university and industry—between academic research and its commercial
application—is a vital cycle. It
should be used more, even during sabbatical years.
I continued my involvement with Teknekron for another quarter century. The ability to simultaneously live both inside and outside the University's walls was a great boon in my professional life. But I leave that tale for another time.
I continued my involvement with Teknekron for another quarter century. The ability to simultaneously live both inside and outside the University's walls was a great boon in my professional life. But I leave that tale for another time.