Joint rate splitting and beamforming design for RSMA-enabled STAR-RIS-assisted ISAC system

In this paper, we consider a rate splitting multiple access (RSMA)-enabled simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS)-assisted integrated sensing and communication (ISAC) system. The meticulously pre-coded common stream in the RMSA scheme not only can ef...

Full description

Saved in:
Bibliographic Details
Published in:Digital signal processing Vol. 162; p. 105146
Main Authors: Chen, Yutan, Wang, Ze, Mo, Minghui, Xu, Hongbo, Zhou, Aizhi
Format: Journal Article
Language:English
Published: Elsevier Inc 01.07.2025
Subjects:
Deep unfolding
Integrated sensing and communication
Rate splitting multiple access
Simultaneously transmitting and reflecting reconfigurable intelligent surface
Deep unfolding
Integrated sensing and communication
Rate splitting multiple access
Simultaneously transmitting and reflecting reconfigurable intelligent surface
ISSN:1051-2004
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, we consider a rate splitting multiple access (RSMA)-enabled simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS)-assisted integrated sensing and communication (ISAC) system. The meticulously pre-coded common stream in the RMSA scheme not only can effectively substitute individual radar sequences to satisfy radar sensing requirements and reduce interference between dual functions, but also mitigate interference among users. Specifically, this paper aims to maximize the achievable sum rate by jointly optimizing the common stream rate allocation vector, the active beamforming matrix at the base station (BS), and the passive beamforming matrix at the STAR-RIS on the premise of ensuring the communication quality of service (QoS) constraint users, the total common stream rate, the transmit power budget, the physical characteristics of STAR-RIS elements, and the minimum signal-to-interference-plus-noise ratio (SINR) required by sensing. Due to the non-convex constraints and objective function, as well as the existence of highly coupled variables, the optimization problem is challenging. Given the impressive capabilities of deep neural network (DNN) in function approximation and faster convergence speed, a deep unfolding (DU) algorithm based on iterative gradient descent (GD) algorithm is proposed. Specifically, the GD algorithm is unfolded into a multi-layer iterative structure, and in each layer, trainable iterative parameters are introduced to speed up the convergence. However, DNN is not suitable for solving constrained optimization problems, so we use Lagrange dual transformation to transform the original problem into the Lagrange dual problem. Then we use the proposed DU algorithm to solve this problem. The numerical simulation results show that the proposed DU algorithm exhibits fast convergence, and compared to the non-orthogonal multiple access (NOMA) scheme, the RSMA scheme demonstrates significant advantages in maximizing the achievable sum rate.
ISSN:1051-2004
DOI:10.1016/j.dsp.2025.105146