SAFETY-CRITICAL INTERNET CONNECTIVITY

[James Pascoe]

Dear All,

I am new to this newsgroup so please forgive me if this is a daft posting, but,
I am interested in looking at network connectivity and real-time control of
safety-critical networks over a temporally unstable network.

Has anybody come across anything that may be useful? The reason I ask is because
I am thinking about doing my Phd in this area and would like to get an insight
into whether or not a lot has been done in this area.

Thanks for your help,

James

--
**************************************************************************
* James Pascoe (research student) | Tel: +44 118 9875123 Ext 4223 / 7626 *
* Department of Computer Science | Fax: +44 118 9751994 *
* The University of Reading | *
* RG6 6AY | Email: J.S.P...@reading.ac.uk *
* United Kingdom | Ja...@james-pascoe.com *
* ---------------------------------------------------------------------- *
* Official URL: http://www.cs.reading.ac.uk/cs/people/jsp/home.html *
* Personal URL: http://www.james-pascoe.com/ *
**************************************************************************

[David Emrich]

James Pascoe <J.S.P...@reading.ac.uk> wrote in message
news:38981605...@reading.ac.uk...

> Dear All,
>
> I am new to this newsgroup so please forgive me if this is a daft posting,

[snip]

Not at all, it's a serious distributed computing problem! And real-time,
mission/safety-critical, fault tolerant, distributed computing networks
(whether over the Internet or not) are about as hard as it gets!

You could probably find a wealth of info in a good distributed computing
text book or thesis.

I believe there are numerous well known and well studied problems about
"fault tolerance", "safety critical-ness" and other related issues (or
similar names for the same thing) that have and are still being researched
(and I believe there are some unsolved ones too), in the distributed
computing environment.

A well designed fault-tolerant distributed system has to assume that
ANYTHING can fail, either temporarily or permanently, and that includes the
data and/or control information paths between the nodes, just as importantly
as the nodes themselves. The problem is exacerbated a bit more when you do
some kind of election process to remove a faulty node (which isn't the only
way to do fault-tolerance, but it is an often-used way). If there are a
bunch of nodes that would swing the vote a different way, and they miss the
chance to "vote", an otherwise good node might be removed from the network
at the whim of a few errant nodes that can co-operate. (Sounds very
conspiratorial doesn't it! Maybe I'm over-anthropomorphising it, but I've
seen some faulty systems that you'd swear were "out to get you"!).

Good luck with the thesis, you could make it really big if you solve one of
the unsolved "mysteries" in this field.

David.

[Niall Murphy]

On Wed, 02 Feb 2000 11:33:25 +0000, James Pascoe
<J.S.P...@reading.ac.uk> wrote:

>Dear All,
>

>I am new to this newsgroup so please forgive me if this is a daft posting, but,
>I am interested in looking at network connectivity and real-time control of
>safety-critical networks over a temporally unstable network.
>
>Has anybody come across anything that may be useful? The reason I ask is because
>I am thinking about doing my Phd in this area and would like to get an insight
>into whether or not a lot has been done in this area.
>

I am not sure how useful it would be, but a lot of fault-tolerant
issues have been addressed (though not necessarily solved) in the
CORBA world. I do not have any exact references though.

good luck ,
Niall

See the User Interfaces for Embedded Systems Page
at http://www.panelsoft.com