Regenerator placement in elastic optical networks with adaptive modulation and coding
Gamgam, Onur Berkay
MetadataShow full item record
Please cite this item using this persistent URLhttp://hdl.handle.net/11693/32214
Due to the rapid and diverse increase in the tra c load on the optical networks, e cient utilization of the network resources becomes an important issue. Using di erent modulation formats and coding rates in optical signal transmission, it is possible to assign di erent spectral e ciency and optical reach for each tra c requests. To satisfy the quality of transmission (QoT) for the distances beyond optical reach, optical - electronic - optical (O/E/O) 3R regeneration of the optical signal is required. During the regeneration process, the spectral e ciency and thus optical reach of the resultant signal can also be set. In these circumstances, by selecting speci c regenerator node locations and assigning di erent line rates for each tra c request, the network utilization can be optimized. Joint selection of regenerator placement (RP), routing and adaptive modulation and coding (AMC) pro le in elastic optical networks (EON) is studied to propose an o ine RP algorithm for a given network topology with link length and link capacity constraints. For a given RP, an Integer Linear Programming (ILP) model is formulated to perform routing and AMC pro le assignment for each tra c demand. We use two di erent approaches for determining candidate paths for routing: In the rst set, k shortest paths (KSP) are utilized for all cases. In the second set, namely regenerator location dependent path selection (RLDPS), the candidate paths are determined according to the given RP. To nd the minimum cost RP among all possibilities, Tabu Search based regenerator placement algorithm (TSRPA) is proposed. Results show that adaptively selecting the candidate paths based on the regenerator locations reduces network utilization either by decreasing the number of regenerator nodes by up to 66.6% or decreasing link capacity utilization by up to 5.09% as compared to selecting candidate paths as xed k shortest paths. The regenerator node location distribution obtained with RLDPS concentrated on smaller number of nodes compared to the results obtained with KSP. By placing regenerators at a signi cantly less number of nodes, capital expenditures (CAPEX) are reduced by RLDPS.