Deep learning-assisted adaptive modulation level assignment for video communication over an elastic optical network

Abstract

Abstract In this paper, we present a user-oriented adaptive modulation level assignment scheme for an elastic optical network (EON) with a focus on video streaming services. Our objective is to maximize total spectral efficiency providing the user quality-of-experience (QoE). The proposed adaptive modulation level assignment for video communication has three stages: network quality-of-transmission (QoT) estimation, extracting the received video quality based on a proposed utility function and finally applying deep learning to select an appropriate modulation level that guarantees user QoE with minimum bandwidth utilization. Simulation results show the efficiency of this approach in increasing the spectral efficiency of the network, especially for long source-destination distances. This increase in EON spectral efficiency leads to a remarkable decrease in blocking probability of the network.