Joint Source-Channel Coding and Bayesian Message Passing Detection for Grant-Free Radio Access in IoT

TL;DR

This paper proposes a joint source-channel coding approach with Bayesian message passing detection for grant-free IoT radio access, enabling direct event detection without individual sensor decoding, improving efficiency over traditional methods.

Contribution

It introduces a non-orthogonal TBMA-based JSC coding scheme and a Bayesian detection algorithm for IoT, enhancing event detection efficiency in grant-free access.

Findings

Bayesian message passing outperforms conventional detection methods.

Joint source-channel coding reduces decoding complexity.

Proposed scheme achieves higher detection accuracy.

Abstract

Consider an Internet-of-Things (IoT) system that monitors a number of multi-valued events through multiple sensors sharing the same bandwidth. Each sensor measures data correlated to one or more events, and communicates to the fusion center at a base station using grant-free random access whenever the corresponding event is active. The base station aims at detecting the active events, and, for each active event, to determine a scalar value describing each active event's state. A conventional solution based on Separate Source-Channel (SSC) coding would use a separate codebook for each sensor and decode the sensors' transmitted packets at the base station in order to subsequently carry out events' detection. In contrast, this paper considers a potentially more efficient solution based on Joint Source-Channel (JSC) coding via a non-orthogonal generalization of Type-Based Multiple Access (TBMA). Accordingly, all sensors measuring the same event share the same codebook (with non-orthogonal codewords), and the base station directly detects the events' values without first performing individual decoding for each sensor. A novel Bayesian message-passing detection scheme is developed for the proposed TBMA-based protocol, and its performance is compared to conventional solutions.