Physically Consistent Evaluation of Commonly Used Near-Field Models

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.

💡 Research Summary

The paper “Physically Consistent Evaluation of Commonly Used Near-Field Models” presents a critical assessment of simplified models prevalent in near-field multi-antenna communication research and introduces a comprehensive, physics-based alternative. The authors argue that while interest in near-field communication is surging, most existing literature relies on overly simplistic models (primarily spherical-wave approximations) whose validity for real-world systems with complex antennas and reconfigurable intelligent surfaces (RIS) is questionable.

The core contribution is the development of a “physically consistent sampled near-field model” for general reconfigurable electromagnetic structures (REMS), encompassing conventional antenna arrays and RIS. This model decomposes the system into three subsystems: the RF frontend (power amplifiers), a reconfigurable tuning network, and the radiating structure. Using scattering parameters (S-parameters) and circuit-theoretic power waves, the model establishes a linear input-output relationship between the amplifier voltages (and incident waves) and the resulting electromagnetic field at a discrete set of points in the region of interest. A key practical feature is that its parameters can be extracted from real-world measurements or full-wave electromagnetic simulations (e.g., using Ansys HFSS), bridging the gap between abstract theory and practical implementation. The model’s output is the electromagnetic energy density, serving as a proxy for received power without the complications of near-field receiver coupling.

Armed with this robust evaluation tool, the authors conduct a systematic comparison against the commonly used spherical-wave model across three test scenarios. In Scenario I (ULA beamfocusing in free space), the spherical-wave model performs adequately for basic beamfocusing and even when applying tapering to suppress sidelobes. However, its prediction error grows significantly when evaluated across a range of frequencies, as it fails to capture the narrowband frequency response of the practical patch antennas used.

Scenario II introduces a dielectric obstacle (a sphere with human-average material properties) between the ULA and the focal point. The spherical-wave model, treating the obstacle as a perfect absorber blocking line-of-sight paths, can still focus energy at the target location behind the obstacle. Nevertheless, its predictive accuracy severely degrades in the shadow region surrounding the obstacle. In contrast, the physically consistent model, which inherently accounts for scattering, diffraction, and reflection from the obstacle, achieves superior beamfocusing and accurately predicts the entire field distribution.

Scenario III evaluates an RIS reflecting an incident plane wave to a near-field focal point. Here, the limitations of the spherical-wave model become starkly apparent. While it manages to focus energy at the correct coordinate, it fails to accurately predict the surrounding sidelobe structure. This inaccuracy stems from its inability to model specular reflections from the RIS surface. Furthermore, similar to Scenario I, its predictions become unreliable when the operating frequency deviates from the center frequency due to its lack of frequency-dependent component modeling.

The paper’s conclusive finding is that while simple spherical-wave models are sufficient for basic conceptual understanding and initial beamfocusing design in ideal free-space conditions, they are insufficient for advanced system design. Their shortcomings in predicting sidelobes, frequency dependence, and performance in complex environments with obstacles or reflective surfaces can lead to significant performance overestimation and suboptimal system configuration. Therefore, for high-precision near-field communication, sensing, and control applications, the adoption of physically consistent models that incorporate mutual coupling, element patterns, frequency response, and environmental interactions—as demonstrated by the proposed framework—is essential.