Distance Adaptive Shared Path Protection for Elastic Optical Networks under Dynamic Traffic

Abstract: Recently, the internet traffic demand has been compoundly rising up as a result of the increase in the number of users as well as data demand per user. That is why, Elastic Optical Networks (EONs), which employ Orthongonal Frequency Division Multiplexing (OFDM) , have been proposed to scale the demands by efficiently utilizing the spectrum as they provide finer spectrum granularity and distance adaptive modulation formatting. Not only efficiency and scalability but also survivability of the network is significant since even a single-link failure may cause huge volume of data considering that even a channel bandwidth may vary between 1 Gb/s and 1Tb/s. Hence, we propose a heuristic algorithm to increase the spectrum efficiency in EONs employing Shared Path Protection (SPP) as the recovery scheme provided that the traffic demand is dynamic and the modulation format is distance adaptive. Our algorithm, Primary First-Fit Modified Backup Last-Fit (PF-MBL), follows two step approach for Routing and Spectrum Assignment (RSA). In the first step, k-shortest path algorithm is applied and candidates paths are found regardless of spectrum availability for routing. In the second step, spectrum is assigned to working paths and backup paths starting from the different ends of the links’ frequency domain so as to group working and backup path resources separately. In working path spectrum assignment, First-Fit strategy is employed. In backup path spectrum assignment, the algorithm chooses a path according to a formula among candidate paths with available spectrum widths found by Last-Fit strategy. In this manner, we expect to provide less fragmented spectrum for backup paths as well as the network, thereby increasing their sharability and thus the spectrum efficiency. We compare our algorithm and the two current solutions by simulations. Results show that PF-MBL can improve the performance in terms of blocking and bandwidth blocking probability by 24% up to 59% compared to the current outperforming algorithm when the bandwidth acceptance ratio of the system varies from 90% to 99.9% in different loads. Moreover, it achieves between 41% to 59% savings over the current outperforming algorithm when the bandwidth acceptance ratio of the system varies from 99% to 99.9%.