Sum-Rate Optimisation of a Multi-User STAR-RIS-Aided System with Low Complexity

Sadaf Syed, Wolfgang Utschick, Michael Joham · 2025

Read Sum-Rate Optimisation of a Multi-User STAR-RIS-Aided System with Low Complexity by Sadaf Syed, Wolfgang Utschick, Michael Joham (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-19723.txt.

Introduction

Syed, Utschick, and Joham tackle a design problem for future wireless networks, in a paper published in 2025: smart surfaces that both reflect and pass radio signals. A plain smart surface only bounces signals to one side, limiting coverage. The dual-mode surface serves users on both sides at once. Prior work tuned such systems by recasting the problem for heavyweight solver software, which runs far too slowly for real networks.

The authors propose a low-complexity method to maximize the total data rate across many users. Their approach avoids the slow generic solvers while keeping performance high. Fast tuning is what makes the surfaces usable outside simulations. Base stations must re-aim these surfaces in near real time as users move. The paper brings the dual-mode surface closer to practical deployment in next-generation networks.

Catalog record

CREATORSadaf Syed, Wolfgang Utschick, Michael Joham
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE30 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-19723

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Sum-Rate Optimisation of a Multi-User STAR-RIS-Aided System with Low Complexity Sadaf Syed, Wolfgang Utschick, Michael Joham Abstract Reconfigurable intelligent surface (RIS) is a promising technology for future wireless communication systems. However, the conventional RIS can only reflect the incident signal. Hence, it provides a limited coverage, as compared to a simultaneously transmitting and reflecting RIS (STAR-RIS). Prior works on the STAR-RIS address the power minimisation or the sum-rate maximisation problem by reformulating the objective problem as a convex optimisation problem and then employing numerical tools like CVX to obtain the solution, which introduces significant computational complexity leading to a huge runtime, making the algorithms impractical for real-world implementation. In this paper, we propose a low complexity solution for the optimisation of a multi-user STAR-RIS system, where the non-convex optimisation problem is decomposed into multiple convex sub-problems with closed-form optimal solutions. The simulation results illustrate that our proposed algorithm achieves similar performance to CVX-based solutions in the literature while being computationally efficient. I Introduction Reconfigurable intelligent surface (RIS), which is a prospective low-cost technology, has received considerable attention for B5G/6G systems [ 1 , 2 , 3 , 4 , 5 , 6 , 7 ] . Specifically, an RIS is composed of a large number of reflecting elements, where the phase shifts can be configured to achieve the desired system’s performance. RIS consists of a large number of lossless reconfigurable reflecting elements that can reflect the incident signal in a controlled manner to achieve the desired performance objectives. Additionally, the energy and hardware costs of the lossless elements of the RIS are significantly lower compared to the traditional active antennas at the base station (BS). Thus, they can be scaled much more easily than the antennas at the BS. Most existing algorithms consider reflecting-only RIS, which can only reflect the incident signals [ 1 , 2 , 3 , 4 , 5 , 6 , 7 ] . This means that the conventional RIS can only serve the users located on the same side as the BS. This geographical constraint limits the flexibility and performance gains, as the users are typically located on both sides of the RIS. To overcome this limitation, the concept of simultaneously transmitting and reflecting RIS (STAR-RIS) was proposed in [ 8 , 9 , 10 ] , where the incident wireless signal on each element of the STAR-RIS is split into reflecting and transmitting components. The transmitted and reflected signals can be reconfigured via independent coefficients by manipulating the electric and magnetic currents of each element [ 9 ] . Compared to the conventional RISs, STAR-RIS requires optimisation of the amplitude coefficients of each element in addition to the phase shifts. The amplitude coefficients determine the percentage of energy allocated to reflection and transmission modes, which are coupled due to energy conservation laws. This makes the optimisation problem of STAR-RIS very challenging. The existing works on STAR-RIS like [ 8 , 10 ] employ interior-point solvers like CVX [ 11 ] to obtain the optimal passive beamforming vectors of the STAR-RIS. However, this results in high computational complexity and runtime, making the algorithms infeasible for real-world implementation. In this work, we firstly reformulate the problem to a tractable form using the fractional programming (FP) approach [ 12 , 13 ] . We propose a low complexity iterative solution where the phase shifts and the amplitude coefficients of the STAR-RIS are solved in closed-form, which leads to much lower computational complexity and smaller runtime as compared to the CVX-based algorithms of [ 8 , 10 ] . II System Model We consider the downlink (DL) of a multi-user STAR-RIS-aided system where the BS is equipped with M antennas and the STAR-RIS consists of N lossless elements. The wireless signal from the BS that is incident on a single element of the STAR-RIS is divided into a transmitting and a reflecting signal. We consider K single-antenna users where K_{\text{r}} users lie on the reflecting side and K_{\text{t}} users lie on the transmitting side of the STAR-RIS, such that K_{\text{r}}+K_{\text{t}}=K . For each element of the STAR-RIS, the phase shifts for reflection and transmission can be designed independently [ 9 ] . However, the amplitude coefficients of the reflection and transmission parts are coupled by the law of energy conservation since the elements of the STAR-RIS are lossless. The reflection and transmission coefficient matrices of the STAR-RIS are given by \displaystyle{\boldsymbol{\it{\Phi}}}_{\text{r}} \displaystyle=\text{diag}\left(\sqrt{\rho_{\text{r},1}}\theta_{\text{r},1},\cdots,\sqrt{\rho_{\text{r},N}}\theta_{\text{r},N}\right) (1) \displaystyle{\boldsymbol{\it{\Phi}}}_{\text{t}} \displaystyle=\text{diag}\left(\sqrt{\rho_{\text{t},1}}\theta_{\text{t},1},\cdots,\sqrt{\rho_{\text{t},N}}\theta_{\text{t},N}\right) (2) where the phase shifts for the reflection and transmission for the n -th element are denoted by \theta_{\text{r},n} and \theta_{\text{t},n} respectively, with \sqrt{\rho_{\text{r},n}} and \sqrt{\rho_{\text{t},n}} being the corresponding amplitude coefficients with \rho_{\text{r},n},\>\>\rho_{\text{t},n}\in[0,1] . Additionally, we have |\theta_{\text{r}/\text{t},n}|=1\>\forall\>n , and \boldsymbol{\varphi}_{\text{r}/\text{t}}=\mathrm{diag}({\boldsymbol{\it{\Phi}}}_{\text{r}/\text{t}}) denotes the corresponding vector. Moreover, we have the constraint that the energy of the incident signal must be equal to the sum of the energies of the reflecting and transmitting signals, i.e., \displaystyle\rho_{\text{r},n}+\rho_{\text{t},n}=1\>\forall\>n. (3) As it can be seen from ( 3 ), one can control the amount of signal reflected and transmitted by each element of the STAR-RIS by adjusting the amplitude coefficients \rho_{\text{r},n}\>\>\text{and}\>\>\rho_{\text{t},n} . Three operating protocols for STAR-RIS have been proposed in the literature [ 8 , 10 ] . 1. Energy Splitting (ES): In this protocol, all elements of the STAR-RIS simultaneously operate in reflecting and transmitting modes with \rho_{\text{r},n},\>\>\rho_{\text{t},n}\in[0,1] and \rho_{\text{r},n}+\rho_{\text{t},n}=1\>\forall\>n . 2. Mode Splitting (MS): In this configuration, some elements of the STAR-RIS fully operate in the reflecting mode and the others fully operate in the transmitting mode, i.e., \rho_{\text{r},n},\>\>\rho_{\text{t},n}\in\{0,1\} . If N_{\text{r}} denotes the number of elements operating in the reflecting mode with \rho_{\text{r}}=1 , \rho_{\text{t}}=0 , and N_{\text{t}} denotes the number of elements operating in the transmitting mode with \rho_{\text{r}}=0 , \rho_{\text{t}}=1 , we have N_{\text{r}}+N_{\text{t}}=N . 3. Time Splitting (TS): In this protocol, all elements of the STAR-RIS alternatingly operate in full transmitting or reflecting modes in different time slots. If \delta\in[0,1] denotes the percentage of time allocated for the reflecting mode and 1-\delta is the fraction of time for the transmitted mode, the algorithm needs to optimise \delta to achieve the optimal performance. The coupling of the reflecting and transmitting parts due to the amplitude coefficients makes the optimisation of ES and MS modes of STAR-RIS complicated. The TS mode is easier to optimise because of the decoupled reflecting and transmitting modes in each time slot. However, it requires strict time synchronisation, which makes its hardware implementation more complicated. Additionally, since the TS mode does not serve all the users in the same time slot, its gain is much lower compared to the ES and MS modes where all the users are simultaneously served. It is to be noted here that the MS protocol is a subset of the ES protocol containing only the binary values of the amplitude coefficients. Hence, the MS protocol cannot achieve the similar gain as the ES protocol, and so, we restrict our discussion to ES and TS modes in this paper. The direct channel from the BS to the k -th user is denoted by {\boldsymbol{h}}_{\mathrm{d},k}\in\mathbb{C}^{M\times 1} , that from the RIS to the k -th user is denoted by {\boldsymbol{r}}_{k}\in\mathbb{C}^{N\times 1} and the channel from the BS to the RIS is denoted by {\boldsymbol{T}} .

Read the complete text