[SystemSafety] Statistical Assessment of SW ......

[Peter Bernard Ladkin]

On 2015-01-24 00:12 , Matthew Squair wrote:
> Another non-trivial hardware problem is how to ensure a shared concept of time in a distributed
> system in the presence of clock drift. Said drift can lead to quite different responses to inputs
> from redundant identical components, based on their hitting a time gate at slightly different moments. 

I thought Leslie Lamport and Fred Schneider essentially solved this problem thirty years ago. Are
you saying the HW people haven't get got it on board? The original problem came from the SIFT
verification attempt at SRI, along with the Byzantine Generals problem.

See for example item #62 on Lamport's publications list, a JACM paper:
http://research.microsoft.com/en-us/um/people/lamport/pubs/pubs.html#clocks

Schneider, for example: https://www.cs.cornell.edu/fbs/publications/byzClockBandol.ps

Here's a survey paper from 1990 from Barbara Simons, Jennifer Welch and Nancy Lynch which I didn't
know about until I looked: http://groups.csail.mit.edu/tds/papers/Lynch/lncs90-asilomar.pdf

Ulrich Schmid did a lot of work on it in the 1990's, but I don't know it well. Here's a keynote talk
he gave in 2007 https://ti.tuwien.ac.at/ecs/people/schmid/Mypapers/ispcs.pdf and a list of selected
papers, which is unhelpfully presented without titles and with a somewhat inefficient window
generation: https://ti.tuwien.ac.at/ecs/people/schmid/Mypapers

PBL


Prof. Peter Bernard Ladkin, Faculty of Technology, University of Bielefeld, 33594 Bielefeld, Germany
Je suis Charlie
Tel+msg +49 (0)521 880 7319  www.rvs.uni-bielefeld.de