[an error occurred while processing this directive]
[an error occurred while processing this directive]
2002-2003 Seminar Series
|
Architectural and Algorithmic Issues in the Design of TDM Grooming
Fabrics
|
Carl McCrosky
Principal Engineer
PMC - Sierra, Inc.
Saskatoon, Saskatchewan
|
DEPARTMENT SEMINAR
|
DATE:
|
Monday, September 23, 2002
|
|
TIME:
|
3:30pm
|
|
PLACE:
|
Engineering 2C40
|
|
|
*** Everyone is welcome ***
|
|
Abstract
Sometimes the simplest problems are remarkably hard. The switch and
grooming of time division multiplexed (TDM) signals is a good example.
Industry needs switching fabrics for TDM protocols such as SONET
(Synchronous Optical Network); these fabrics serve to route
sub-component signals to the destinations, to optimally pack the TDM
carriers, and to support automatic protection switching. There are
two principal ways to build such fabrics: as strictly non-blocking
single stage fabrics, and as multistage fabrics based on the ideas of
Clos and time-space-time switching. We'd all build SNB fabrics if we
could, but it's very difficult to realize large SNB fabrics. So much
of the industry has focused on multistage fabrics. The problem is
that these fabrics require complex open path algorithms (OPA), and yet
they can still block (fail to make connections from unused inputs to
unused outputs). This presentation will explore this problem domain,
consider the related practical issues, and propose several new ways to
build SNB fabrics.
About the speaker
Carl has wandered just about every which way in his career of more
than 30 years. He's worked in the software industry He's done CS
and EE post-graduate degrees (Queen's University, Kingston). He
worked as an RA at MIT as part of his PhD. He's worked in the telecom
hardware industry.
His interests range from functional programming languages, to
representations and algorithms for Boolean functions, to VLSI systems,
to hardware synthesis, to switching algorithms and fabrics, and to
automatic protection for telecommunications systems. Carl wrote a
mess of academic papers during his years as a Professor of Computer
Science at the University of Saskatchewan - a couple of those papers
are still worth reading. Carl supervised about 50 graduate students
while working as a professor - every one of those students has
become a highly valuable worker somewhere in the information
technology industry. Carl misses teaching.
[an error occurred while processing this directive]