DeepNP: Deep Learning-Based Noise Prediction for Ultra-Reliable Low-Latency Communications

Abstract: Closing the gap between high data rates and low delay in real-time streaming applications is a major challenge in advanced communication systems. While adaptive network coding schemes have the potential of balancing rate and delay in real-time, they often rely on prediction of the channel behavior. In practice, such prediction is based on delayed feedback, making it difficult to acquire causally, particularly when the underlying channel model is unknown. In this work, we propose a deep learning-based noise prediction (DeepNP) algorithm, which augments the recently proposed adaptive and causal random linear network coding scheme with a dedicated deep neural network, that learns to carry out noise prediction from data. This neural augmentation is utilized to maximize the throughput while minimizing in-order delivery delay of the network coding scheme, and operate in a channel-model-agnostic manner. We numerically show that performance can dramatically increase by the learned prediction of the channel noise rate. In particular, we demonstrate that DeepNP gains up to a factor of four in mean and maximum delay and a factor two in throughput compared with statistic-based network coding approaches.