Self-learning monitoring on-demand strategy for optical networks

In current dynamic optical networks with cascaded filters and amplifiers, the optical signal-to-noise ratio (OSNR) can vary significantly from channel to channel. Under such uncertainty, OSNR prediction for unestab-lished channels becomes indispensable but remains a big challenge. For protective network planning purposes such as margin threshold setting or wavelength assignment, it is desirable to evaluate the worst OSNR performance of each network link. Such exploration will unavoidably employ active monitoring probes, which may cause interruptions to the network. An efficient active monitoring strategy that optimizes the choice of probes or monitoring trials is needed. We propose a "self-learning" monitoring strategy integrated at intermediate nodes. Our method can intelligently select the channel to be monitored in order to search for the target global maxima of OSNR degradation for a specific link. Our monitoring scheme detects intermediate node OSNR in the linear regime. It is shown that our method can predict the target OSNR value with only 0.5 dB error while reducing the required monitoring data by up to 91% compared to conventional methods.