Integrated Sensing and Communications for Unsourced Random Access: A Spectrum Sharing Compressive Sensing Approach

TL;DR

The paper tackles unsourced random access in an ISAC context for massive machine-type communication by proposing SSCS-UNISAC, a spectrum-sharing compressive sensing framework. It introduces separate sensing and communication codebooks, a power-split parameter $eta$, and a joint non-linear decoder that operates with ADCE, segment stitching, SISO decoding, and SIC, plus MUSIC-based AoA estimation for sensing. The approach yields significant capacity gains (20–30 dB in many scenarios) over conventional protocols like TDMA and ALOHA, and remains robust as the number of active users or antennas grows. This work advances practical UNISAC design by integrating CS-based multi-user detection with joint sensing capabilities, enabling efficient spectrum sharing and scalable mMTC support.

Abstract

This paper addresses the unsourced/uncoordinated random access problem in an integrated sensing and communications (ISAC) system, with a focus on uplink multiple access code design. Recent theoretical advancements highlight that an ISAC system will be overwhelmed by the increasing number of active devices, driven by the growth of massive machine-type communication (mMTC). To meet the demands of future mMTC network, fundamental solutions are required that ensure robust capacity while maintaining favorable energy and spectral efficiency. One promising approach to support emerging massive connectivity is the development of systems based on the unsourced ISAC (UNISAC) framework. This paper proposes a spectrum-sharing compressive sensing-based UNISAC (SSCS-UNISAC) and offers insights into the practical design of UNISAC multiple access codes. In this framework, both communication signals (data transmission) and sensing signals (e.g., radar echoes) overlap within finite channel uses and are transmitted via the proposed UNISAC protocol. The proposed decoder exhibits robust performance, providing 20-30 dB capacity gains compared to conventional protocols such as TDMA and ALOHA. Numerical results validate the promising performance of the proposed scheme.

Figures (6)

• Figure 1: Illustration of uplink transmission in UNISAC system.
• Figure 2: Illustration of the proposed UNISAC encoder and decoder designs. Notably, the encoder design for sensing does not imply that the SUs are coordinated or intend to transmit data. The sensing signals can be equivalently treated as simultaneous radar echoes from multiple targets.
• Figure 3: Illustration of stitching segments by channel estimation: Each selected candidate at next node is determined by evaluating the resemblance of channel estimation via \ref{['eq:6']}, i.e., The channel coefficients can be treated as natural user tags due to the random nature of the channel conditions.
• Figure 4: Illustration of error rate under different power allocation ratio $\beta$ with $E/N_0=20$dB, $|\mathcal{A}_c|+|\mathcal{A}_s|=200$, $M=5$.
• Figure 5: Minimum-required energy-per-user $E/N_0$ (dB) with a small number of receiving antennas, $M=5$. The benchmarks ISAC-URA include the theoretically achievable and optimistic (performance floor) bounds of UNISAC, along with conventional protocols (ALOHA and TMDA-MUSIC).