Abstract
The architecture of next-generation high-speed networks will likely consist of
IP and WDM layers. The IP layer runs packet switching, whereas the WDM layer uses
circuit switching. Although the two share the goal of efficient use of network
resources, there is a lack harmony between the two approaches, owing to different
traffic-engineering capabilities. To this end, we introduce a combined approach for
quantitatively analyzing the effectiveness of traffic engineering for the two
layers. First, we model an edge router and devise a traffic-engineering mechanism to
minimize the combined cost over a single path. We show that both the explicit price,
which characterizes the traffic-engineering cost, and the implicit price, which
determines the network congestion cost, play a key role in wavelength allocation. We
consider wavelength granularity, which significantly affects overall performance
when the number of wavelengths in a link is small. To solve this granularity
problem, we propose an approximation algorithm that finds an integer solution from
the real solution. Through simulations, we verify that our proposed algorithm
achieves near-optimal performance with greatly reduced complexity and ensures the
feasibility of our proposed solutions.
© 2004 Optical Society of America
PDF Article
More Like This
Cited By
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Contact your librarian or system administrator
or
Login to access Optica Member Subscription