I. Introduction
With the recent deployment of 5G cellular services, addressing the measurement complexity and capacity needs of Radio Base Stations (RBS) has become a major topic within the wireless industry. On the regulatory side, RBS have to comply with radio frequency (RF) electromagnetic field (EMF) exposure regulations. EMF compliance testing aims at determining the compliance boundary or exclusion zone, outside which the exposure is below the applicable limits. For macro RBS, evaluation is typically carried out using field strength measurements or computations. For low-power RBS, the compliance boundary might extend only a few centimeters away from the product and compliance can be assessed by means of Specific Absorption Rate (SAR) measurements. SAR assessment traditionally involves a 6-axis robot arm with a single detected probe, scanning the volume of a body- simulating mannequin (or phantom) [1]. For RBS, whole-body SAR measurements, which require field probing over a large volume, might be needed. As it is extremely time-consuming, faster alternative techniques are of high interest, especially as the number of test modes is multiplying with 5G. This paper introduces such an approach, based on the combination of antenna or Over- The-Air (OTA) measurements, near-field equivalent source reconstruction and full-wave electromagnetic (FWEM) simulation. Previous publications have exhibited related hybrid solutions, also converging experimental and numerical methods, e.g. [2], [3]. Yet, to our knowledge, this paper is the first to demonstrate an effective association of OTA and FWEM modelling for SAR evaluation purposes. This new approach raises three fundamental questions, which are partially investigated and answered in this publication: (i) as equivalent currents are reconstructed from measurements taken at larger range length, some reactive near-field components are not captured. Is this loss of information problematic for the accuracy of the evaluation of local and spatially-averaged SAR metrics? (ii) how well is the actual source-phantom interaction modelled, when solving the problem of a free space field source in presence of a virtual mannequin with a Finite Difference Time Domain (FDTD) code? (iii) is the technique applicable to testing of commercial devices, where no RF port is accessible and wide-band digitally modulated signals are transmitted?