A Real-Time OSNR Penalty Estimator Engine in the Presence of Cascaded WSS Filters

Virtual Optical Networks (VONs) offer scientists the opportunity to test and validate Software Defined Networking (SDN) controllers without requiring access to expensive optical equipment. A key component in VON is the real-time optical layer emulation (OLE) engine that must estimate the OSNR at the virtual receiver of every virtual lightpath that is established in the VON. This study investigates two algorithms for training a two-hidden layer Neural Network (NN) whose aim is to estimate the OSNR penalties caused by a series of virtual Wavelength Selective Switches (WSS) that are traversed by a virtual lightpath before reaching its virtual receiver. The NN solution yields good accuracy and only requires a few microseconds of computation time — a two key requirements for achieving a meaningful.