Interference Estimation via Model-Based Deep Learning in Grant-Free Networks

Abstract

This letter presents a novel approach for estimating interference distribution parameters in grant-free access networks using a model-based deep learning (DL) framework. Our method integrates the precision of analytical models with the adaptability of deep learning algorithms. Specifically, we employ an analytical model to generate labeled data, which enhances the deep learning model’s ability to estimate interference levels. Through extensive validation, we demonstrate that our approach accurately estimates interference across a broad range of scenarios, including operating regions not covered during the model’s training. Moreover, our method also estimates the spatial density of interfering nodes, making it a valuable tool for interference management in grant-free access networks. This methodology offers a robust solution for improving interference estimation accuracy, aiding decision-making at the Medium Access Control (MAC) and physical layers in grant-free access schemes.