[Paper Review] Empirical Evaluation of a 28 GHz Antenna Array on a 5G Mobile Phone Using a Body Phantom

Lauri Vähä-Savo, Christian Cziezerski+5 more|arXiv (Cornell University)|Sep 14, 2020

Antenna Design and Analysis33 references22 citations

TL;DR

This paper presents a hexagonal physical human body phantom for repeatable empirical evaluation of 28 GHz millimeter-wave antenna arrays on 5G mobile phones. By comparing measured realized gain patterns with numerical simulations using a realistic human model, the study validates that the phantom accurately reproduces spherical coverage statistics, demonstrating <1 dB variation across repeated measurements and confirming its suitability for robust antenna array testing under realistic body effects.

ABSTRACT

Implementation of an antenna array on a 5G mobile phone chassis is crucial in ensuring the radio link quality especially at millimeter-waves. However, we generally lack the ability to design antennas under practical operational conditions involving body effects of a mobile user in a repeatable manner. We developed numerical and physical phantoms of a human body for evaluation of mobile handset antennas at 28 GHz. While the numerical phantom retains a realistic and accurate body shape, our physical phantom has much simpler hexagonal cross-section to represent a body. Gains of the phased antenna array configuration on a mobile phone chassis, called co-located array is numerically and experimentally evaluated. The array is formed by placing two sets of 4-element dual-polarized patch antenna arrays, called two modules, at two locations of a mobile phone chassis. Modules are intended to collect the maximum amount of energy to the single transceiver chain. Spherical coverage of the realized gain by the array shows that the experimental statistics of the realized gains across entire solid angles agree with numerical simulations. We thereby demonstrate that our antenna evaluation method reproduces the reality and our phantom serves repeatable tests of antenna array prototypes at 28 GHz.

Motivation & Objective

• To address the lack of repeatable, standardized methods for evaluating 5G mmWave antenna performance under realistic human body effects.
• To develop a physical body phantom with a simplified hexagonal cross-section that accurately replicates electromagnetic scattering from a human body at 28 GHz.
• To validate the phantom’s accuracy by comparing experimental measurements with numerical simulations using a realistic human model.
• To demonstrate the repeatability and reliability of antenna array measurements using the physical phantom in both free space and body-impacted conditions.
• To establish a practical test framework for device conformance testing and comparative evaluation of antenna array designs at mmWave frequencies.

Proposed method

• Development of a numerical human body model in CST Microwave Studio with surface impedance to represent skin at 28 GHz, using FTDT and surface integral equation solvers.
• Creation of a physical hexagonal body phantom with a dielectric material layer to emulate human skin, designed to match the electromagnetic scattering of the numerical model.
• Implementation of a co-located 2×4-element dual-polarized patch antenna array on a mobile phone-sized chassis, with two modules placed on opposite sides to maximize spatial diversity.
• Use of equivalent near-field sources to decouple the simulation of antenna feeds and body interaction, enabling accurate scattering field computation.
• Far-field measurement setup with a calibrated probe antenna, using a 3D measurement tower to capture spherical coverage of realized gain across 2048 synthesized beam patterns.
• De-embedding of feed line losses from measured data to enable direct comparison with simulated gains.

Experimental results

Research questions

• RQ1 Can a simplified hexagonal physical body phantom accurately reproduce the spherical coverage of realized gain patterns from a realistic human body at 28 GHz?
• RQ2 How well do experimental measurements using the physical phantom match numerical simulations using a full 3D human model?
• RQ3 What is the repeatability of antenna array measurements under the influence of the physical body phantom?
• RQ4 How do measurement uncertainties—such as tower shadowing and phase center misalignment—affect the accuracy of spherical coverage statistics?
• RQ5 To what extent does the presence of the body phantom affect the maximum realized gain and coverage performance of the phased array?

Key findings

• The measured spherical coverage CDF of the antenna array matches the simulated CDF above the 0.7 probability level, with only a 2.2 dB difference at the 0.1 outage level in free space, primarily due to absorber use in measurements.
• With the body phantom present, the median realized gain remains above 4.5 dBi, and the measured and simulated CDFs differ by only 0.3 dB at peak gain and 3 dB at the 0.1 outage level.
• Two repeated measurements with the body phantom show less than 1 dB difference across the entire CDF, confirming high measurement repeatability.
• The physical phantom causes a slightly wider shadowing region (≈7° wider) around φ = 270° compared to the numerical model, likely due to dimensional differences in the phantom's width.
• Measured gains near θ = 135° and θ = ±180° are lower than simulated due to antenna misplacement and tower shadowing, respectively, contributing to up to 0.6 dB amplitude uncertainty.
• The study confirms that the hexagonal body phantom enables repeatable, accurate, and reliable evaluation of mmWave antenna arrays, with good agreement between simulation and experiment.