Global Internet traffic is expected to increase fourfold in the next five years. This traffic will be highly time-varying and has led to the introduction of flexible, so-called *elastic,* optical networks. In such networks, light-paths are set up and optically routed from sources to destinations with no intermediate electro-optic conversions. Moreover, signal parameters such as carrier spacing, signaling power, bit and code rates, and modulation formats can be adjusted in real time, based on the traffic demand and network load. This requires a network control layer that dynamically updates light-paths, spectrum allocations, and other parameters, and a link control layer that provides input to the network control layer and adapts to its control signals. This proposal deals with the design and analysis of these two control layers. For a single wavelength channel in a point-to-point link, coherent reception and digital signal processing enable operation close to the nonlinear Shannon capacity limit. However, the presence of linear and nonlinear crosstalk from other channels in the same fiber generally forces an operation at far (typically 10 dB) lower power levels. This proposal suggests that improved channel models, signal processing, and network adaptivity will enable operation with significantly reduced margins. The obtained results will enable a large-scale optimization and significantly increased throughput of reconfigurable optical networks in a nonlinear environment.
The project is closed: 31/12/2016