Deep-Learned Approximate Message Passing for Asynchronous Massive Connectivity

TL;DR

This paper introduces a deep learning-based algorithm called LAMP for efficient joint user activity detection, delay estimation, and channel estimation in asynchronous massive connectivity systems, outperforming traditional methods.

Contribution

It proposes a novel model-driven deep learning approach, LAMP, tailored for asynchronous massive connectivity, reducing complexity and improving robustness over traditional compressed sensing algorithms.

Findings

LAMP significantly outperforms conventional AMP in simulations.

LAMP demonstrates robustness to delay spread in asynchronous users.

Three LAMP structures balance performance and complexity in multi-antenna scenarios.

Abstract

This paper considers the massive connectivity problem in an asynchronous grant-free random access system, where a huge number of devices sporadically transmit data to a base station (BS) with imperfect synchronization. The goal is to design algorithms for joint user activity detection, delay detection, and channel estimation. By exploiting the sparsity on both user activity and delays, we formulate a hierarchical sparse signal recovery problem in both the single-antenna and the multiple-antenna scenarios. While traditional compressed sensing algorithms can be applied to these problems, they suffer high computational complexity and often require the perfect statistical information of channel and devices. This paper solves these problems by designing the Learned Approximate Message Passing (LAMP) network, which belongs to model-driven deep learning approaches and ensures efficient performance without tremendous training data. Particularly, in the multiple-antenna scenario, we design three different LAMP structures, namely, distributed, centralized and hybrid ones, to balance the performance and complexity. Simulation results demonstrate that the proposed LAMP networks can significantly outperform the conventional AMP method thanks to their ability of parameter learning. It is also shown that LAMP has robust performance to the maximal delay spread of the asynchronous users.