Routing and Spectrum Assignment in Elastic Optical Networks integrating Quality of Transmission requirements
针对弹性光网络中的路由与频谱分配问题，提出用线性约束近似计算光信噪比，从而在整数线性规划模型中集成传输质量要求，实验验证该方法能高效得到满足工程需求的优化解。
In Elastic Optical Networks (EONs), the frequency spectrum is divided into narrow frequency slots, and sequences of contiguous slots form channels. Given an EON and a set of traffic demands, the NP-hard Routing and Spectrum Assignment (RSA) problem consists of establishing for each traffic demand a lightpath, composed of a source-to-destination path and a channel, such that no two traffic demands share a frequency slot on any optical fiber along their paths. Besides this non-overlapping constraint, the paths also need to fulfill some Chromatic Dispersion (CD) and certain Optical Signal to Noise Ratio (OSNR) requirements to guarantee the end-to-end Quality of Transmission (QoT). While the CD requirements can be easily integrated as linear constraints in any Integer Linear Programming (ILP) model for the RSA problem, the handling of OSNR requirements is more involved due to the nonlinear nature of the propagation of the optical signals along the fibers of the network. To compute the OSNR of uncompensated transmission paths in optical coherent transmission systems, the Gaussian Noise model, involving some nonlinear computations, is commonly used by network operators to assess their operational validity. Our contribution is to integrate the OSNR computation with the help of linear constraints that can be easily integrated in any ILP model for the RSA problem. We empirically verify that our linear constraints well approximate the OSNR computed using widely adopted tools. Hence computed lightpaths satisfy all necessary operational requirements. • Quality of Transmission metrics are critical for optical networking optimization. • Quality of Transmission estimation can be accurately modeled through simple equations. • Optimization produces optimal solutions that fulfill network engineering requirements. • Integrating Quality of Transmission reduces computational time and improves solutions.