Intelligent Transport Systems - Telenor

Nils Jacob Berland (41)
received his PhD (Dr.Scient.) in
Computer Science in 1993. In
addition to his work for Telenor
R&D in the area of security he
has also worked with scenariobased
planning and business
development in other units of
Telenor. In 1998 he started the
consulting company Pantarei
AS where he is still employed.
He holds positions as part-time
associate professor at the Norwegian
University of Science
and Technology, Trondheim,
and at Molde University College.
His professional interests are
distributed systems and optimisation,
and control systems
for smart homes.
njb@pantarei.no
Telektronikk 1.2003
Designing PRT Control Systems by
Internet Protocol Thinking –
Comments on Ed Anderson’s Paper
NILS JACOB BERLAND
In his paper from 1998, reprinted in this issue of Telektronikk, Ed Anderson presents his design criteria
for a PRT control system. For most of the period in which the idea of PRT has developed, the computing
and networking technology for building a PRT control system were less mature and hence the
thinking related to the design of control system was naturally affected by this. As no PRT system for
other than restricted test tasks has so far been set into operation, there still does not exist any readymade
software and hardware for controlling a PRT system. Further, there are no standards, neither for
operating such a system, nor for solving the interconnect problems that arise if two or more such
networks were to exchange vehicles.
In this short comment we look for similarities between some of the PRT proposals and the Internet
Protocol. This similarity may be exploitable when designing and building an actual PRT control system.
Specifically, it is a goal to set the issue of PRT in the context of telecommunication. We outline how a
control system for PRT, meeting Anderson’s design criteria, could actually be implemented by adapting
the thinking, huge competence and technology base existing for IP networks. What we propose is not
new to the PRT designers, but hopefully the IP way of thinking can bridge two professional and business
cultures that have or at least should have a lot in common.
However, the IP networks metaphor is not the only way to design a PRT system. We will also briefly
touch an alternative to the IP metaphor. Needless to say – the alternative also has parallels in existing
designs of telecommunication systems.
Distributed Control System
Thinking
The thinking on software control systems for
PRT started many years ago, when IT systems
were thought of in more centralised terms and
implemented with heavier building blocks and
fewer commodity products than today. Some
PRT designs and implementations are thus fairly
parallel to designs of computer and telecom networks
from the same period, e.g. being based on
full synchronisation of the network and time
slots, with a central control system allocating
time slots to each vehicle. Later came various
modifications, implying an increasing degree of
distribution of computing power and decision
making competence to the various basic elements
of the system: vehicles, stations and intersections.
(An overview of such principles is
found in the PhD thesis of Markus Theodor Szillat:
A Low-level PRT Microsimulation, PhD dissertation,
University of Bristol, April 2001.)
The evolution towards distributed control is also
reflected in Ed Anderson’s paper. Each vehicle,
station and intersection in a PRT system is supposed
to be unmanned and more or less make
its own decisions related to routes and operation.
This requires a distributed control system combined
with mobile computing and intensive communication.
Distributed control systems have been around
for years and have been the domain of telecom
operators and a few other big companies or government
agencies. Generally such systems have
been expensive, but now cheaper computers and
open standards for distributed computing make
them more feasible. Consider the case of air traffic
control. Air traffic control has until now
relied on ground control of airspace, and aircraft
following strict corridors and time slots. However,
now the United States Federal Aviation
Authority has proposed the concept of “free
flight” where the task of separation assurance is
delegated to each aircraft. It is even argued that
“free flight” will reduce costs and improve efficiency
and capacity.
Mobile computing has also been around for
years – NASA’s spacecraft may be the prime
example. Demand for mobile computing based
on small computers connected through some
kind of radio network has more or less exploded,
and devices complete with software for radio
based connectivity through GSM, WLAN or
Bluetooth are ubiquitous. NASA’s mobile computing
was certainly not affordable, but now
almost everyone can buy mobile computing
equipment.
One of the factors that drive the development of
mobile computing and distributed control systems
– aside from the demand for services – is
the availability of Internet Protocols and appropriate
tools. IP and networks have transformed
the way distributed systems are designed. This
will probably be the case for PRT systems too:
The first control system to meet Anderson’s
design criteria will in all probability be based
117