Efficient optical amplification in self-healing synthetic ROADMs

Optical backbone networks are carrying enormous amount of traffic and therefore, network reliability performance is extremely important to minimize the service interruption time and loss of data. Besides, increasing energy consumption of the ICT sector makes network energy efficiency gain a lot of attention. Reducing the usage of components in the network is a promising strategy to both make network more energy efficient and reliable as it allows (i) putting idle components into a power-saving, or sleep, mode and (ii) reusing them as redundancy for failure recovery. The new generation of synthetic reconfigurable optical add-drop multiplexers (ROADMs) implemented by Architecture on Demand (AoD) supports the above functionalities by offering unprecedented flexibility and self-healing capabilities. In AoD nodes, components are interconnected via high-port-count optical switch serving as optical backplane, and each connection uses only the components necessary to fulfill the switching and processing requirements. In this paper, we aim at improving optical network availability and reducing power consumption by minimizing the number of optical amplifiers used in AoD nodes. To this end, a decision on the necessity of signal amplification is made for every connection according to the power level at each node. Unnecessary amplifiers are bypassed and put into sleep mode, decreasing the overall power consumption of the network. This also reduces the associated risk of connection failure, while reuse of idle amplifiers for failure recovery further improves connection availability. To gain greater insight into the benefits of our approach, we analyze the performance of the NSF and EON network deploying AoD nodes with different optical backplane implementations, i.e. 3D MEMS and piezoelectric optical switch (POS), and compare them with networks deploying hard-wired ROADMs. Simulation results show that deployment of synthetic AoD nodes with POS as optical backplane can reduce the network mean down time and power consumption of EDFAs by up to 63% and 38%, respectively, on average over all test cases.