Spatially-Coupled Random Access on Graphs

TL;DR

This paper introduces a spatially-coupled protocol for coded slotted ALOHA that leverages LDPC code principles to significantly improve random access performance, approaching theoretical limits.

Contribution

It proposes a novel spatially-coupled access protocol inspired by LDPC convolutional codes, achieving near-optimal performance in random access scenarios.

Findings

Attains performance close to the theoretical limits of CSA

Demonstrates threshold saturation effect in the proposed protocol

Allows high traffic throughput close to 1 packet per slot

Abstract

In this paper we investigate the effect of spatial coupling applied to the recently-proposed coded slotted ALOHA (CSA) random access protocol. Thanks to the bridge between the graphical model describing the iterative interference cancelation process of CSA over the random access frame and the erasure recovery process of low-density parity-check (LDPC) codes over the binary erasure channel (BEC), we propose an access protocol which is inspired by the convolutional LDPC code construction. The proposed protocol exploits the terminations of its graphical model to achieve the spatial coupling effect, attaining performance close to the theoretical limits of CSA. As for the convolutional LDPC code case, large iterative decoding thresholds are obtained by simply increasing the density of the graph. We show that the threshold saturation effect takes place by defining a suitable counterpart of the maximum-a-posteriori decoding threshold of spatially-coupled LDPC code ensembles. In the asymptotic setting, the proposed scheme allows sustaining a traffic close to 1 [packets/slot].