Abstract
Deep learning (DL) finds rich applications in the wireless domain to improve spectrum awareness. Typically, the DL models are either randomly initialized following a statistical distribution or pretrained on tasks from other data domains such as computer vision (in the form of transfer learning) without accounting for the unique characteristics of wireless signals. Self-supervised learning enables the learning of useful representations from Radio Frequency (RF) signals themselves even when only limited training data samples with labels are available. We present the first self-supervised RF signal representation learning model and apply it to the automatic modulation recognition (AMR) task by specifically formulating a set of transformations to capture the wireless signal characteristics. We show that the sample efficiency (the number of labeled samples required to achieve a certain accuracy performance) of AMR can be significantly increased (almost an order of magnitude) by learning signal representations with self-supervised learning. This translates to substantial time and cost savings. Furthermore, self-supervised learning increases the model accuracy compared to the state-of-the-art DL methods and maintains high accuracy even when a small set of training data samples is used.
Abstract (translated)
URL
https://arxiv.org/abs/2207.03046
