Physically Consistent Evaluation of Commonly Used Near-Field Models

TL;DR

The paper tackles the problem of validating widely used near-field models in multi-antenna and RIS-enabled systems. It introduces a physically consistent sampled near-field REMS framework that predicts the electromagnetic field at discrete coordinates, with parameters obtainable from measurements or full-wave simulations. Through three beamfocusing scenarios, the study finds that traditional spherical-wave-based approaches suffice for basic focusing but fail to predict sidelobes and frequency-dependent behavior, particularly for RISs where specular reflections are important. The results emphasize the need for physically grounded near-field modeling in design and analysis, and the authors provide open-source code to enable reproducible evaluation. Overall, the work offers a rigorous, coordinate-based near-field modeling approach that improves accuracy for complex REMS configurations and practical RIS applications.

Abstract

Near-field multi-antenna wireless communication has attracted growing research interest in recent years. Despite this development, most of the current literature on antennas and reflecting structures relies on simplified models, whose validity for real systems remains unclear. In this paper, we introduce a physically consistent near-field model, which we use to evaluate commonly used models. Our results indicate that common models are sufficient for basic beamfocusing, but fail to accurately predict the sidelobes and frequency dependence of reflecting structures.

Figures (3)

• Figure 1: (a) Structure of the proposed model: The RF frontend comprises $N$ power amplifiers, the tuning network represents the reconfigurable RF circuitry, and the radiating structure consists of $M$ antenna elements. The inputs to the model are the voltages of the power amplifiers and incoming plane waves; the output is the electromagnetic field at a prespecified discrete set of coordinates. (b) Beamfocusing example from sec:beam_focus_scenario, showing the sampled energy density $u(\,r,\theta,\varphi)$ in the region of interest.
• Figure 2: Ansys HFSS screenshots showing a patch-antenna element and a RIS unit cell. The ULA elements are excited via a coaxial feed; the RIS unit cells feature two ports (in light blue) instead of the varactor diodes.
• Figure 3: Results for Scenarios i@, ii@, and iii@ as summarized in tab:overview. The figures compare the energy density prediction of a commonly used spherical-wave-based model with the prediction of our physically consistent model. The objective in each scenario is to focus the energy density towards a specific focus point. In Scenario i@, the spherical-wave model shows good agreement in the beamforming and tapering cases. However, across different frequencies, deviations increase as the offset from the center frequency grows. In Scenario ii@, the spherical-wave model is able to focus at the target behind the obstacle. However, the predictive accuracy degrades in the region behind the obstacle. In contrast, by employing our proposed physically consistent model, we achieve superior beamfocusing performance. In Scenario iii@, while the spherical-wave model focuses energy on the correct target coordinate, it fails to accurately predict the surrounding sidelobes. Furthermore, the commonly used spherical-wave model does not account for specular reflections from the RIS, and, if the frequency deviates from the center frequency, the spherical-wave model is unable to accurately predict the energy density.