Network restoration using recurrent neural networks

In this paper, a method is proposed for network restoration using a centralized, static restoration after failure, where the restoration initiated at the local node or at the source using a hybrid strategy. The traditional approaches to network (re)routing are based on the heuristics that are prone to enter local minima. A new rerouting algorithm is proposed that attempts to identify the (near-)optimal route based on a recurrent neural network that uses simulated annealing. The proposed method for restoration is compared with the ones based on source, local and local destination based rerouting models.