20.11.2012 Views

Contents Telektronikk - Telenor

Contents Telektronikk - Telenor

Contents Telektronikk - Telenor

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

50<br />

The life and work of Conny Palm<br />

– some personal comments and experiences<br />

BY ROLF B HAUGEN<br />

This paper is an introduction to the life<br />

and works of the great Swedish teletraffic<br />

theoretician Conny Palm. The<br />

focus is on how I personally have used<br />

and experienced Palm’s work throughout<br />

the years rather than trying to<br />

make an exhausted review. Through<br />

this flash back of my years of work in<br />

teletraffic, it is unavoidable not to<br />

mention a few other teletraffic researchers<br />

that came into “my way”.<br />

Nevertheless, the theme throughout<br />

the paper is always Palm and his work.<br />

1 Introduction<br />

The science of Teletraffic theory, founded<br />

by Erlang in the beginning of this century,<br />

has always had a strong foothold in<br />

Scandinavia. The field was first dominated<br />

by the Danes, then gradually Swedish<br />

groups became dominant, until finally,<br />

on the threshold of the seventies, a relatively<br />

strong build-up started in Norway.<br />

By that time professor Arne Myskja and<br />

his colleagues in Trondheim were engaged<br />

in traffic measurements with subsequent<br />

analyses and several interesting<br />

studies of subscriber behavior. At<br />

Kjeller, Richard Solem was busy building<br />

up a scientific group at Norwegian<br />

Telecom Research (NTR) with a rather<br />

broad focus. The teletraffic group at<br />

NTR, the T-group, was soon to be a force<br />

centre in the Norwegian community, initiating<br />

activities in Trondheim as well as<br />

numerous places in Oslo, like the Norwegian<br />

Computer Centre, STK, EB, other<br />

departments of Norwegian Telecom, to<br />

mention but a few.<br />

Personally, I had the privilege to join the<br />

Kjeller group in the middle of the seventies<br />

after some years as ‘traveling’ physicist.<br />

Coming from the University of Oslo<br />

rather than from NTH as did the others in<br />

the group, I was perhaps a bit of an outsider.<br />

However, since we were all physicists<br />

of one kind or another, I quickly fell<br />

in with the professional as well as the<br />

social activities in a fairly acceptable<br />

way. And the atmosphere and activities<br />

in the group were just incredible! It is<br />

rare in any profession to meet so many<br />

extraordinary gifted people in one and<br />

the same group. The T-group soon<br />

became a concept at the institute. Practical<br />

jokes were numerous and legendary,<br />

and in social activities like sports, games,<br />

partying, etc., the challenge at NTR was<br />

always: the T-group versus the rest! But<br />

the professional flag was nevertheless<br />

kept high!<br />

Although my main fields of study were<br />

physics and mathematics, I embarked on<br />

a course in statistics, quite by impulse, at<br />

one point during my studies. By chance, I<br />

chose the subject Probability theory with<br />

a text book written by William Feller, a<br />

person unknown to me at that time. But<br />

seldom have I been more fascinated! It<br />

was obvious that Feller had practical<br />

experience in the field of probability; his<br />

examples and explanations were related<br />

to real life situations. And his approach<br />

and mathematical treatment of the problem<br />

was done in such an intuitive way<br />

that we hardly realized that we, in fact,<br />

were working with the fundamentals of<br />

the underlying theory. Even as a student I<br />

grasped the freshness of his treatment<br />

and was stricken by his pedagogic ability<br />

in explaining intricate problems in a<br />

simple way.<br />

Feller brought me into the world of<br />

stochastic processes, a world fascinating<br />

in itself. But equally fascinating was<br />

Feller’s treatment. I certainly remember<br />

being introduced to the state equations,<br />

the so called birth and death processes<br />

based on equilibrium considerations.<br />

Feller also introduced me to the generating<br />

functions, a powerful means of solving<br />

discrete processes. It was incredible<br />

with which elegance and virtuosity Feller<br />

solved intricate probability problems<br />

with the help of these functions! At least,<br />

so it seemed to me at that stage of my<br />

education.<br />

Hardly could I then know that later, and<br />

for a long period of my life, I was to have<br />

the above mentioned processes and solution<br />

techniques as my daily work. I<br />

remember vaguely that I was introduced<br />

to the names Erlang and Palm; Erlang<br />

being the person behind the state equations<br />

that became the origin of queuing<br />

theory, and Palm one of the pioneers in<br />

establishing the mathematical fundament<br />

of those theories. I also seem to remember<br />

that some of the examples and solutions<br />

in the text book went back to Palm.<br />

Even if I at that time had no inclination<br />

towards telecommunications, I certainly<br />

grasped that Erlang had to do with telephony.<br />

Palm’s background, though, was<br />

more diffuse to me.<br />

Another eight years were to pass before I<br />

again ‘met’ with the two mentioned pioneers.<br />

I had then started at NTR, read the<br />

fundamentals of traffic theory and joined<br />

a project team analyzing the telephone<br />

network of Oslo, the so-called ARONproject.<br />

This project was a collaboration<br />

between STK and NT, represented by<br />

NTR, Technical Department and Region<br />

Oslo. One of the activity leaders was<br />

Eliot Jensen, then at STK, who for me<br />

became like a tutor in the teletraffic area.<br />

The strange thing was that the very first<br />

technical paper I was presented to by<br />

Eliot in this project was, in fact, an article<br />

by Palm: Calcul exact de la perte dans<br />

les groupes de circuits échelonnés [1].<br />

Here, Palm discusses the solution of a<br />

simple grading with the help of state<br />

equations. No closed form solution was<br />

given by Palm, and I set out to solve the<br />

problem by generating functions. It turned<br />

out, though, that no general closed<br />

form could be found by this technique,<br />

either. Hence, I ended up solving the<br />

whole thing numerically on a computer.<br />

This possibility was definitely not easy at<br />

hand in the thirties when Palm looked<br />

upon the problem!<br />

2 Conny Palm<br />

Conny Palm was one of the most brilliant<br />

scientists in the history of teletraffic<br />

theory. Or perhaps I should say, in the<br />

history of queuing theory, since Palm<br />

really laid the mathematical fundamentals<br />

of this theory. He was born in 1907,<br />

joined the Royal Institute of Technology<br />

in Stockholm in 1925, finished, in practice,<br />

his studies in 1931 but did not take<br />

his final exam in electrical engineering<br />

until 1940. For nine years he was thus<br />

what we would call an eternal student!<br />

However, his nine extra student years<br />

were by no means lazy years. From his<br />

publication list we find seven papers<br />

dated 1940 or earlier. From 1934 he<br />

worked within the area of telephony and<br />

joined Ericsson in 1936. He was an<br />

active participant at Cramér’s seminars<br />

on mathematical statistics at Stockholm’s<br />

University where he in 1937 met William<br />

Feller. This meeting turned out to be an<br />

important one for further development of<br />

the theory of stochastic processes; it was<br />

a meeting between an academic scholar<br />

searching for the fundamentals of probability<br />

theory and a ‘practical’ engineer<br />

who happened to have many of the<br />

answers. Palm had obtained these<br />

answers through his own work and the<br />

work of his predecessors within telephony.<br />

According to Cramér: “Palm presented<br />

a unique, mature and clarified<br />

view on his ideas of probability theory”.<br />

And in fact, in 1950 Feller himself wrote:<br />

“Waiting time and trunking problems for<br />

telephone exchanges were studied long<br />

before the theory of stochastic processes<br />

was available and had a stimulating

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!