Optical computer networks provide the means for reliable data communications with low transmission and propagation delay, low error rate, and at the same time, high throughput. Their adoption rate constantly increases for fulfilling connectivity needs in the backbone of large counties or the constant need for access from the end-user perspective. The increase of resource usage leads to higher energy consumption which is a factor that should be confronted by embedding energy-efficiency in the

more »

... putational logic while transferring data. Energy-efficient algorithms design the logical backbone topology of the network with low energy consumption and the preservation of performance at a high level. On the other hand, optical access networks intermediate between end-user units and the backbone, covering the last mile of connectivity. The fast access in these networks from the end-user perspective, improves their quality of service and the overall throughput. Artificially intelligent algorithms exploit the operating state of the network by adapting to its runtime conditions, allocating resources more efficiently than the static alternatives. Thus, being suitable for either the backbone or the access part of the network. Two types of AI algorithms are designed and evaluated which are capable of achieving the aforementioned goals, i.e., those with energy-efficiency in their computational logic for the backbone, and also those offering adaptive bandwidth distribution at the access side of the network.