Convex Channel Power Optimization in Nonlinear WDM Systems Using Gaussian Noise Model

Abstract

Optimization of channel powers to maximize minimum margin or total capacity in WDM systems is studied. Using a Gaussian noise nonlinearity model, the signal-to-noise ratio (SNR) in each channel is expressed as a convex function of the channel powers. Using the SNR expression, convex optimization problems with objectives of maximizing the minimum channel margin or maximizing the fiber capacity minus a coding cap are formulated. Performance gains from software-based power optimization are observed in mesh networks and in point-to-point links having heterogeneous SNR requirements. By contrast, in systems with uniform amplifier noise and modulation formats, the optimized power allocation provides very little improvement over a traditional flat power allocation. In the 14-node NSFNET network, a margin gain of 1.5 dB on average is achieved through power optimization, as compared to a flat power allocation. Margin gains averaging 1.4 dB are found for subsets of this network with three to 13 nodes.