A Markov-Based Reservation Algorithm for Wavelength Assignment in All-Optical Networks

Most routing and wavelength assignment algorithms for all-optical networks do not consider the potential problem of reservation connection, which occurs when two or more reservation requests compete for the same wavelength. Confliction can happen even if the network is only lightly loaded. In this paper, we propose a new reservation protocol Markov-based backward reservation (MBR) based on Markov modeling of the network traffic on optical links for resolving wavelength reservation confliction. Our simulations show that this new protocol can improve network performance as measured by decreased blocking probability. The MBR algorithm works best in small-hop-count networks with a comparatively small number of wavelengths per fiber compared to other frequently considered algorithms.