Comparative Analysis of Ray Tracing and Rayleigh Fading Models for Distributed MIMO Systems in Industrial Environments

Aymen Jaziri 1 , David Demmer 2 , Yoann Corre 1 , Jean-Baptiste Doré 2 , Didier Le Ruyet 3 , Hmaied Shaiek 3 , Pascal Chevalier 3 · 2025

Read Comparative Analysis of Ray Tracing and Rayleigh Fading Models for Distributed MIMO Systems in Industrial Environments by Aymen Jaziri 1 , David Demmer 2 , Yoann Corre 1 , Jean-Baptiste Doré 2 , Didier Le Ruyet 3 , Hmaied Shaiek 3 , Pascal Chevalier 3 (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-2503-01300.txt.

Introduction

Aymen Jaziri and six co-authors compare two ways to model radio channels for distributed MIMO inside a factory. Distributed antennas can beat one central array if user gear stays near several access points. The team used a real 3D factory map to test coverage and capacity. Ray tracing reads the building in detail and keeps correlations true, while stochastic Rayleigh models run faster and generalize.

Downlink and uplink single-user capacity were scored under several deployment layouts. Densifying the access point grid mattered most for the distributed gains. The paper weighs when ray tracing's accuracy is worth its cost against the cheaper stochastic route.

Catalog record

CREATORAymen Jaziri 1 , David Demmer 2 , Yoann Corre 1 , Jean-Baptiste Doré 2 , Didier Le Ruyet 3 , Hmaied Shaiek 3 , Pascal Chevalier 3
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE24 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-01300

Browse the full library · About the Tower of Babel

Full text

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

Comparative Analysis of Ray Tracing and Rayleigh Fading Models for Distributed MIMO Systems in Industrial Environments Aymen Jaziri 1 , David Demmer 2 , Yoann Corre 1 , Jean-Baptiste Doré 2 , Didier Le Ruyet 3 , Hmaied Shaiek 3 , Pascal Chevalier 3 Abstract This paper presents a detailed analysis of coverage in a factory environment using realistic 3D map data to evaluate the benefits of Distributed MIMO (D-MIMO) over colocalized approach. Our study emphasizes the importance of network densification in enhancing D-MIMO performance, ensuring that User Equipment (UE) remains within range of multiple Access Points (APs). To assess MIMO capacity, we compare two propagation channel models: ray tracing and stochastic. While ray tracing provides accurate predictions by considering environmental details and consistent correlations within the MIMO response, stochastic models offer a more generalized and efficient approach. The analysis outlines the strengths and limitations of each model when applied to the simulation of the downlink (DL) and uplink (UL) single-user capacity in various D-MIMO deployment scenarios. AP Access Point UE User Equipment DL downlink UL uplink RU Radio Unit CU Central Unit DU Distributed Unit ULA Uniform Linear Array CCDF Complementary Cumulative Distribution Function CDF Cumulative Distribution Function SU Single-User MU Multi-User MIMO Multiple Input Multiple Output SISO Single Input Single Output D-MIMO Distributed Multiple Input Multiple Output TDD Time Division Duplexing OFDM Orthogonal Frequency Division Multiplexing CSI Channel State Information RSRP Reference Received Signal Power ZF Zero forcing SVD Singular Value decomposition SNR Signal-to-Noise Ratio SINR Signal-to-Interference-plus-Noise Ratio EE Energy Efficiency SE Spectral Efficiency LoS Line-of-Sight NLoS Non Line-of-Sight PE Power Efficiency RT Ray Tracing I Introduction The introduction of private mobile networks is anticipated to revolutionize various industries, including factory automation, transportation and logistics sectors, significantly influencing their operations. Alongside advancements like Mobile Edge Computing (MEC), Time-Sensitive Networking (TSN), and improved security measures, D-MIMO ( D-MIMO ) technology is also expected to enhance communication availability and reliability in indoor settings, such as factories. This is achieved by increasing LoS ( LoS ) coverage and lowering the chances of signal obstruction, as multiple AP can serve individual users. In this paper, we characterize the MIMO propagation channel and present a detailed analysis of the impact of network densification on coverage overlapping and optical visibility. This is accomplished through the realistic modeling of 3D map data and RT ( RT ) simulations in a factory setting, leveraging the 3.7 GHz spectrum, which is designed to deliver low-latency, high-capacity private 5G solutions [ 1 ] . Previous studies have underscored the importance of realistic propagation models in achieving reliable wireless communication in industrial settings [ 2 , 3 ] . In this private network scenario, our objective is to evaluate the potential advantages that D-MIMO technology, in conjunction with network densification, can bring to indoor environments. D-MIMO systems rely on multiple AP to serve UE [ 4 ] . Enhanced coverage overlapping and improved optical visibility ensure that UE are within the range of multiple AP , facilitating the effective implementation of D-MIMO . Additionally, we aim to highlight the disparity between a fully RT -based channel model and the Rayleigh-based fading model that is commonly used in theoretical research. The theoretical benefits of D-MIMO and Cell-Free massive MIMO (CF-mMIMO) over conventional co-located MIMO and small non-cooperative base stations have been extensively investigated in the literature. For instance, a comparison between numerous non-cooperative small cells and CF-mMIMO is presented in [ 5 ] . These studies highlight the potential of D-MIMO and CF-mMIMO to provide superior coverage, increased capacity, and enhanced reliability, making them promising solutions for future wireless communication networks. Despite extensive research into D-MIMO communications, few studies have evaluated the combined impact of network densification and D-MIMO on network performance using realistic RT propagation data (e.g. [ 6 ] for mmWave). This work, therefore, contributes to the evaluation of propagation in an industrial environment based on RT tools and assesses system performance by incorporating a physical layer model capable of advanced, state-of-the-art processing, such as UL ( UL ) cooperative detection and DL ( DL ) cooperative beamforming for multi-stream transmissions. This paper is structured as follows. Section II briefly explains the studied scenario and the considered propagation channel models. Section III details the system model, the power allocation and the spectral efficiency computation. Section IV presents the numerical results and analysis. Finally, Section V concludes the paper and gives future perspective of this work. II System model We consider M\times N distributed MIMO ( MIMO ) OFDM ( OFDM ) transmission with N_{c} sub-carriers. The AP and UE operate under a TDD ( TDD ) protocol, which consists of an UL pilot training phase for channel estimation, an UL transmission and a DL transmission. We assume perfect synchronization among AP in terms of time, frequency, and phase. Additionally, channel reciprocity between UL and DL links is considered, enabling the MIMO system to operate in closed-loop mode with CSI ( CSI ) available at the transmitter. The UL channel estimation can thus be used to determine the DL precoding coefficients. The received signal vector \mathbf{y}_{k}\in\mathbb{C}^{N\times 1} at the k -th subcarrier can be written as \mathbf{y}_{k}=\mathbf{W}_{k}\left(\mathbf{H}_{k}\sqrt{\mathbf{P}_{k}}\mathbf{F}_{k}\mathbf{Q}_{k}\mathbf{x}_{k}+\mathbf{n}_{k}\right), (1) where \mathbf{x}_{k}\in\mathbb{C}^{N\times 1} is the transmitted signal, \mathbf{n}_{k}\in\mathbb{C}^{N\times 1} the noise vector. Each element of the noise vector is uncorrelated and follows a normal distribution, specifically \mathcal{N}(0,N_{0}) . \mathbf{W}_{k}\in\mathbb{C}^{N\times N} , \mathbf{H}_{k}\in\mathbb{C}^{M\times N} , \mathbf{P}_{k}\in\mathbb{R}^{M\times M} , \mathbf{F}_{k}\in\mathbb{C}^{M\times N} and \mathbf{Q}_{k}\in\mathbb{C}^{N\times N} are respectively the linear combining at the receiver side, the propagation channel matrix, the per-antenna amplification gains (diagonal matrix), the linear precoding applied at the transmitter side and the per-layer power allocation (diagonal matrix) 1 1 1 As channel aging is not considered in the proposed study, we decided to omit the time indices for sake of simplicity. . In this work, we consider SU ( SU ) multi-layer transmissions meaning that independent data streams are transmitted in parallel. Therefore, the precoding/combining techniques aim at isolating the different independent streams and avoid inter-layer interference. Two strategies are considered for DL : i) the ZF ( ZF ) where \mathbf{F}_{k}\triangleq\mathbf{H}_{k}^{\dagger} 2 2 2 where x^{\dagger} denotes the right Moore-Penrose pseudo inverse of x and \mathbf{W}_{k} is the identity matrix and ii) the SVD ( SVD ) where \mathbf{F}_{k}\triangleq\mathbf{U}_{k}^{H} and \mathbf{W}_{k}\triangleq\mathbf{V}_{k} for \mathbf{H_{k}}=\mathbf{V}^{H}_{k}\mathbf{\Sigma}_{k}\mathbf{U}_{k} . The main difference between the two approaches is the presence of a combiner for SVD . We assume that combiners \mathbf{W}_{k} are either supplied to or estimated by the UE . In this scenario, the number of UE received antennas matches the number of multiplexed MIMO layers. For both techniques, a water-filling algorithm is applied to each carrier. Power allocation is achieved in a way maximizing the achievable capacity [ 7 ] . The proposed study aims at comparing the achieved performance of legacy co-localized and distributed architectures. Indeed, in co-localized scenarios, when the UE are served by a unique AP , M denotes the number of AP antennas. On the other hand, with distributed architectures, M denotes the total numbers of active antennas ( i.e. number of cooperative AP times the number of antenna per AP ). Indeed, we assume centralized processing, meaning that the DL precoding is computed from the CSI of all AP .

Read the complete text