[LINK] Cityrail computer systems down for for hours on Sunday

[Tom Worthington]

At 5/3/2004 10:37 AM, you wrote:
>... Cityrail computers were dead in the water for hours on Sunday ...

See "Sasser eyed over train outage" by Chris Jenkins, MAY 03, 2004 is at 
<http://australianit.news.com.au/articles/0,7204,9455677%5E15331%5E%5Enbv%5E15306-15319,00.html>:

"NSW TRAINS authority RailCorp has sent in software engineers to find the 
source of the outage that left up to 300,000 commuters stranded yesterday, 
saying the new Sasser worm, which has already spawned two variants, is 
being evaluated as a possible cause. ..."

But it seems more likely an ordinary software fault was responsible, rather 
than a worm.  In any case this is a serious failure in a safety related 
system and there is a need for an independent investigation and public 
report before more lives are lost.

Problems with the Metronet Train radio system are documented in both the 
Glenbrook <http://www.waterfallinquiry.com.au/pdf/glenbrook/glend044.pdf> 
and Waterfall 
<http://www.waterfallinquiry.com.au/InterimReport/Chapter5.htm> rail 
accident reports. But little seems to have been done.

It happens I posted an item to aus.rail a few weeks ago suggesting that the 
Sydney rail system could be retrofitted with a digital packet radio system 
at low cost to improve safety: 
<http://groups.google.com.au/groups?selm=cf94a0b1.0404151507.2216f1c9%40posting.google.com>:

---
Newsgroups: aus.rail
Subject: Fit NSW trains with digital radio accident reporting device?
Message-ID: <cf94a0b1.0404151507.2216f1c9 at posting.google.com>

In "A Sense of Place: GPS and Alaska Rail Safety" (GPS World, March 1,
2004 <http://www.gpsworld.com/gpsworld/article/articleDetail.jsp?id=87472>)
Andrew J. Schiestl describes a system using GPS and VHF radio for a
collision avoidance system (CAS). Perhaps this could be enhanced to
automatically report an accident and its the location. This would have
been of use in the NSW Waterfall accident, where reporting and finding
the location of the accident took some time. The Alaskan system
requires relatively little infrastructure by using VHF packet radio.

Some advanced automobile alarms are equipped with GPS and a mobile
phone interface to report their location if stolen. As a byproduct
these can also detect air bag activation and automatically report that
an accident has occurred and the car's location.

The Alaskan rail system uses GPS to locate the train and a digital
packet radio to report it. I suggest the radio could also be equipped
with low cost accelerometers as used in car air bag controllers to
detect a large deceleration, indicating an accident. The digital radio
could then report the position of the train and signal a possible
accident. This could be done very quickly and automatically, in the
event the train crew was incapacitated.

This may seem superfluous, as a control centre monitoring the position
of each train should alert staff to a problem. Also it would be
expected the train crew would report an accident. However, it may be
difficult for the central system to infer the severity of a problem,
and the train crew may be unable to report. An automated alert from
the train showing excessive deceleration would be a less ambiguous
indication of a problem.

Such a system would have been of use in the Waterfall accident, 31
January 2003, as detailed in the "Waterfall Rail Safety Investigation
Final Report" <http://www.transport.nsw.gov.au/waterfall/>. Section
14.3 "Identification of the emergency" describes how the accident was
initially thought to be a track circuit failure. It was approximately
15 minutes before the guard was able to report the derailment and even
then there was difficulty with identifying the accident location for
emergency services.

If the train was equipped with an automated device it could have
reported the accident and its location within a few seconds. Such a
system could continually report the location of the train, report on
the driver's vigilance and act as a remotely scanned datalogger. The
VHF packet radios are relatively low cost (thousands of dollars for a
radio, millions for the network), as they have a long range and
operate on a low data rate. There may even be an existing government
or commercial digital radio infrastructure which could be used.

ps: I am a computer person, not a railway expert, so please excuse me
if this is all obvious (or plain wrong).
---


Tom Worthington FACS tom.worthington at tomw.net.au Ph: 0419 496150
Director, Tomw Communications Pty Ltd ABN: 17 088 714 309
http://www.tomw.net.au PO Box 13, Belconnen ACT 2617