Ciena Corporation
Reinforcement learning for autonomous telecommunications networks

Abstract:

Systems and methods include obtaining performance monitoring data from a network which operates and generates the performance monitoring data which reflects a current state of the network; responsive to a predetermined reward which quantifies minimizing or maximizing an aspect in the network for Reinforcement Learning, causing one or more actions to be performed in the network, wherein the one or more actions are based on the predetermined reward; obtaining updated performance monitoring data from the network subsequent to the one or more actions to determine an updated state of the network; and continuing the causing the one or more actions to be performed in the network based on the current state and the updated state. The causing the one or more actions to be performed in the network is performed to maximize or minimize a total discounted accumulated reward.