Abstract

In this paper the problem of efficient lightpath allocation in a PCE-based architecture is addressed. We propose a novel algorithm that uses the leveraging of lightpath request buffering and prediction at the PCE. Buffering and prediction decrease blocking and delay of requests, respectively. Simulation results in the ARPANet topology show that the algorithm proposed decreases the percentage of blocked requests (between 6% and 28%) compared to an algorithm without the buffering-prediction capability and diminishes delay (up to half) with respect to a strategy that only uses buffering of requests. © 2011 IEEE.