|
A
Personal Perspective on the History of Transfinite Systems Ltd.
By John Parker
When
I first met John Pahl in the second half of 1992, he was working on
a proposal to develop simulation software for Inmarsat. I had been introduced
to John by our manager at the time because I had some experience of
interference analysis and software development, and we were set to work
together on the proposal.
As is often the case, the scope of the Inmarsat requirement seemed to
far outstrip the funds that they had available, and John was in the
process of trying to match the work to the budget.
Looking at the various options proposed my first reaction was that anything
less than the full spec was not going to do the job. This coupled with
my natural, unfettered optimism (my gravestone will carry the words
"How difficult can it be?") lead to a proposal to do the full amount
of work for the original budget.
Nightmare!
A small team worked long days and some weekends to deliver both a functional
piece of software and a preliminary analysis of potential interference
from the proposed Inmarsat Project 21 options.
As I recall the final report and V1 of the software was delivered to
Inmarsat in early 1993.
It was obvious to me at that early stage that John and I had complementary
skills and contrasting approaches that just happened to contribute to
a great team. This was also commented on by our managers and by Kumar
Singarajah who was then the technical officer for the project at Inmarsat.
A break in the Inmarsat development program meant that John and I parted
ways for a while - John continued to work for Inmarsat and I went to
work for the UK Ministry of Defence on land communications and electronic
warfare simulation.
Our thoughts turned to a possible gap in the communications market.
John and I had both noticed that, over a number of years, we had been
working on projects re-inventing the wheel as far as simulation and
interference modelling were concerned. What is the difference between
a radio used on a battlefield, or on a geostationary satellite or a
non-geostationary satellite? The answer is - essentially very little.
Differences between comms systems are a consequence of the way basic
elements are combined and of geometric and dynamic aspects. Was it possible
that a set of building blocks and their interactions could be defined
such that a tool generally useful in simulation could be produced?
During covert meetings in the Windmill Pub on Clapham Common, I must
admit that my first reaction to John's suggestion that this was possible
and could form the basis of a business was, NO!!.
Initially I thought that either it would be too hard, or that someone
else had already done or was doing it. I also doubted that there was
a market, having been involved in the development of a software tool
for GSO Appendix 29 calculations that sold about 6 copies in 10 years.
After some thought I came to a rather different conclusion. A general
tool does not have to do absolutely everything. Provided it does a range
of things well then it is worthwhile and could be re-sold a number of
times.
The above-mentioned GSO tool had never been actively marketed, but still
generated interest, years after any development had finished. This indicated
that people were having problems with analysis that they thought software
could help with.
The satellite communications industry was on the edge of a revolution,
exciting new dynamic systems were being proposed, which would certainly
require simulation tools.
If anyone was able to produce such a tool it was John Pahl.
John being very solicitous by nature had gone further than he revealed
to me initially, and had applied for UK Government funding to develop
the prototype. His proposal for a SMART Award from the DTi had reached
the stage where around 70 proposals were being considered for 35 Awards.
In the end the proposal was rejected on the basis that a software product
is in some way intangible, and would not be an asset, at the end of
the development process.
In the absence of the SMART Award, the decision to quit our badly paid,
un-rewarding jobs was slightly more difficult. Within a month we had
resigned and we were working as Transfinite Systems in October 1994.
The company name means "beyond limits" and had been suggested by John
Pahl, whose background is in mathematics - Transfinite
numbers being those larger than infinity (in a set theoretical formal
framework).
We had some consultancy work which kept us alive financially for a while,
but the main purpose of the company was the development of an interference
modelling tool (provisionally named IMT), and this took most of our
attention.
At around the same time Inmarsat was spawning ICO and the ICO system
was beginning to take shape and to generate an enormous amount of interest
world-wide. We continued to work on ICO proprietary software, and in
parallel, but completely separately began to develop our product.
Naturally when we considered how our product would address MSS sharing
with FS some of our experience with ICO was very useful. However, it
must be clearly stated that no part of our commercial product was derived
from software developed for anyone else.
During the early part of 1995 we had many discussions about a name for
the commercial release of the software, Suggestions including SIGINT,
IMT, SIGNET, names with COMM in them, names with SIM in them, Visual
Interference Tool, Visual this, Visual that. What we wanted to get across
was that we were selling a tool with a graphical interface and graphical
feedback, but that it was not merely a way of producing pretty pictures.
We have always tried to emphasise the underlying analysis capabilities.
I must claim responsibility for merging the two words Visualise and
Analyse to give Visualyse. The name had some advantages...
It can be registered as a trademark because it is not an English word
The Y in the middle gives some scope for graphic designers to play around.
(note that we have not yet taken advantage of this.)
The disadvantage is that no one spells it correctly.
In parallel to the development of the first commercial release, we were
also pursuing a contract for the Radiocommunications Agency in the UK.
The Agency wanted a tool which had a specification very close to the
one we had developed for Visualyse. Our proposal to develop and jointly
market this tool was rejected in favour of a competitive bid. At the
time, I think John Pahl saw this as a major blow, I was less concerned
and in a way slightly relieved to be free to develop the software as
we saw fit. However, we were once again left without direct support
for the development and had to seriously consider the situation.
In the background we had a continuing good relationship with Inmarsat/ICO.
Consultancy and development work for the Spectrum management group in
preparation for WRC 95 and beyond kept Transfinite Systems going through
the year. In fact we found it necessary to take a new team member on
board. Robin Bowes, who initially took up an informal shortish contract,
joined us. He remains in the same role today. I had worked with Robin
on comms simulation development before, and his experience with terrain
and propagation modelling was useful.
Visualyse Version 1 was demonstrated at EUTELSAT in May 95. Though incomplete
as an analysis tool much of the interface was done and the animated
graphics were in place. We thought that feedback before a commercial
release would be useful, and this certainly turned out to be the case.
My thanks to EUTELSAT for this.
The commercial launch of Visualyse Version 2 was on October 18th 1995
in Geneva. The launch coincided with WRC95. Thirty people attended a
product presentation lasting around 1 hour. Comments were very positive,
although no-one handed over purchase orders on the spot.
Following quickly from this first presentation were visits to, among
others, the Radiocommunications Agency, the Civil Aviation Authority
and BT. Sales were made in each case.
A demonstration to COMSAT also resulted in a sale, and suddenly Transfinite
Systems was an international company.
Spurred on by the success of Version 2 we began development of Version
3 almost immediately. Around this time we joined by our software development
manager, Malcolm Barbour.
It is important to understand that, with the exception of some discussions
with EUTELSAT and some experience with ICO software, we had no real
idea if we had produced a tool that people needed or what the demand
for Visualyse would be. Within a few months we were seeing ITU-R Study
Group papers produced using Visualyse, and getting feedback from users.
We realised that people were mainly using the tool to do large, complex
simulations containing many stations and links. Visualyse Version 2
provided some support in the definition of constellations, but when
the number of links became large, simulations could become unmanageable.
Visualyse Version 3 extended and improved Version 2, introducing tools
to manage complex simulations and analyse large quantities of data,
improving graphics and generally updating look ëní feel.
Visualyse Version 3 was announced in December 1996 and released early
in 1997. Sales increased rapidly, and it became clear that many people
had been waiting for the new release. Sales remained strong through
1997, Transfinite Systems became a medium sized company with the recruitment
of new programming staff, office management and administration staff.
Although Visualyse Version 3 was greatly extended, there remained a
number of specialist areas that could not be addressed by the software.
The concept of modular add-ons was introduced and internal discussions
combined with customer feedback, lead to four new modules which would
be available to work with the new release (to be called Visualyse Professional).
The add-on modules were:
Terrain
Traffic
The Define Variable Module
The IDWM Module
The Terrain module introduced new capabilities in modelling terrestrial
links. The development of Terrain included new windows to view path
profiles, new propagation models looking at diffraction over irregular
terrain and new functions to import and manage digital elevation data.
We concentrated initially on the data format used in the development
of the GLOBE data at the NGDC. The GLOBE data promised low resolution,
cheap data covering the whole world. Since the release of the module
we have also worked on other formats and defined a Visualyse internal
format which can be used to import from any source.
The terrain module opened up possibilities for Visualyse in terrestrial
network markets and terrestrial broadcasting. Up to this time Visualyse
had been perceived as a satellite analysis tool, although it has always
been more general than this.
The Traffic module was produced in response to a perceived need to model
variations in load levels on interfering links. Interference analysis
based on maximum loading or even average loading will inevitably be
limited in some way ñ maximum loading will lead to pessimistic
results, average loading may miss peak interference events. The variation
in Link loading is what the Traffic Module was about.
The Define Variable Module began life as the Monte-Carlo module. The
idea was to allow any simulation variable to be defined stochastically
from a user defined distribution. During the course of the development,
a couple of other ideas came up. Firstly, it would be good to be able
to set the value of a variable at each timestep directly, or to set
the value at a intervals and interpolate between these points. Secondly,
how about defining a variable by an offset value from another variable.
Applications of these features were many, including defining routes
for mobile stations, randomising satellite locations for NGSO constellations,
setting transmit power based on received signal level etc etc.
Introducing the new functions made the module more useful, but very
difficult to name and therefore to market. The current name ëDefine
Variableí or DefVar in some places does not really convey the
power and utility of the module. The name Monte Carlo module only tells
half the story.
The IDWM module allows data for some of our propagation models to be
taken from the Integrated Digital World Maps produced by the ITU.
The release of Visualyse Professional and the modules again saw an upturn
in sales, and an increase in interest in the product. Early 1998 saw
the 100th license of Visualyse shipped.
The latter half of 1998 was our most active period to date, and we have
no reason to expect any slowdown in the foreseeable future.
Transfinite Systems is now ready to expand again. New Managing Director
Steve Farmer, expects a doubling of turnover and profit over the next
12 month period, and following a extensive survey of user expectations,
some exciting new developments for Visualyse are in the pipeline.
Key Dates:
Sep 1994
Transfinite Systems Ltd incorporated as a Limited company
Oct 1994
John Pahl and John Parker begin working as partners in Transfinite Systems
Apr 1995
Visualyse Version 1 demonstrated at EUTELSAT in Paris
WRC 95
Visualyse Version 2 launched in Geneva, 13th November 1995
Jan 1996
First Visualyse copy shipped to the Civil Aviation Authority
Dec 1996
Visualyse Version 3 Announced
Mar 1998
Visualyse Terrain. Traffic and Monte Carlo modules released.
|