**** Columbia Network Research Group Seminar ****

Title:    A Quantitative Measure of Topology Life 
          Time for Telecommunications Networks
Speaker:  Nick Maxemchuk 
          AT&T Labs- Research/Columbia University 
When:     Wednesday, November 1, 2000, 10 am
Where:    Interschool Lab, CEPSR Building, 7th floor
          Columbia University

Work with:
   Iradj Ouveysi      Moshe Zukerman
   Telstra Labs       Melbourne University

Perhaps the title of this talk should be "How I spent my winter
vacation."

When I first got to the University of Melbourne last fall the first
truly dense wavelength division multiplexing products were becoming
available, along with a schedule for future improvements.  Moshe and I
spent our time trying to figure out how this would affect networks,
while Iradj, a student, did all of the hard work.

An important effect of the current and future DWDM is that the fibers
that are in place will have adequate bandwidth for a longer period of
time.  If a fiber provided twice as much capacity as was needed in 1995,
and the requirements on the link double every year, in 2007 that fiber
will still have twice as much capacity as is needed, because of DWDM.

In the meanwhile, corporations move their offices, the population ages
and moves to less expensive real estate markets, and Internet server
farms are constructed in Idaho.  The result of having adequate bandwidth
for a longer period of time is that the network will have to survive
more unexpected changes.  The current network design procedures that use
linear growth, or projected shifts in load are no longer adequate.

We have proposed a new measure for network performance evaluation called
topology lifetime.  The measure provides insight into which one of a set
of topologies is likely to last longest before more capacity must be
installed.  The lifetime measure is not single valued, but considers
growth as a function of a set of demand shifts (perturbation).  One
network may be better able to support a uniform growth in the traffic,
while another may support more growth when unexpected shifts in the load
occur.

The lifetime measure is applied to several topologies; an hierarchical
network that is typical of current telecommunications networks, a dual
ring that the SONET people have fallen in love with, and several regular
arrays.  The current telecom structures are designed to support the
current network loads and don't deal well with the unexpected.  Sonet
rings should only be installed in Australia, where the center of the
continent is unoccupied.  Regular arrays respond best to the unexpected. 
These are the structures that we would recommend if we were doing a
scorched earth design.  In the meanwhile, the main use of our tool is to
compare the next fiber installations in a network.  We try to determine
which installation will lead to the longest time before we have to dig
another ditch.

This work will appear in Globecom 2000.  A longer version of the paper,
that has been submitted for journal review is available at
www.research.att.com/~nfm .

Bio:

Nick Maxemchuk got his BSEE from the City College of New York and his
MSEE and PhD from the University of Pennsylvania.

In the late 60's he joined the David Sarnoff Research Center, when RCA
sold 90% of the color TV sets in the world.  He witnessed the demise of
that giant.  In the mid 70's he joined the AT&T Bell Laboratories, when
Ma Bell was a benevolent monopoly.  He's lived through it's divestiture
and division.  Now he coming to Columbia.

Along the way he's written a lot of papers and received a lot of
patents.  He's gotten his share of awards, held adjunct positions at a
couple of universities, and has advised anyone who would listen.  The
details are on his WEB site www.research.att.com/~nfm.