1. Home
  2. Featured Articles
  3. Over the Air Testing for Antenna Arrays: Challenges and Solutions

Over the Air Testing for Antenna Arrays: Challenges and Solutions


by Reiner Stuhlfauth, Technology Marketing Manager, Rohde & Schwarz, Munich and Dr. Corbett Rowell, Technology Marketing Manager, Rohde & Schwarz, Munich

5G will apply multiple antenna systems and combine them with enhanced spatial multiplexing to provide data for multiple users, known as massive MIMO. Despite their advantages, the industry is facing some challenges to characterize such antenna array systems. In the future, it will no longer be possible to do all the performance evaluation of the radiation pattern in a conducted way, so connection over-the-air will be essential. This article presents technical aspects on how to measure three-dimensional antenna patterns using an over-the-air testing setup.

The upcoming standard for future mobile communication systems known as 5G does not only aim at providing services like enhanced mobile broadband (eMBB), ultra-reliable low-latency communications (uRLLC) or massive machine type communications (mMTC), but also at providing more capacity with a much higher flexibility and all together at lower operational expenses (OPEX). Two technologies are being discussed to tackle those goals: on the one hand, virtualization or software defined network SDN, on the other using massive MIMO multiple antenna scenarios. For obtaining a wider bandwidth that guarantees a higher throughput, 5G systems will use higher frequencies in the cm and mmWave ranges. One drawback is the higher free space path loss. According to the well-known equation by Friis, free space path loss depends on the frequency, antenna gain and on a propagation environment specific exponent factor Y

One idea to compensate for this high free space path loss is the usage of antenna arrays that provide a much higher antenna gain due to beamforming (GTX and GRX). Channel analysis have shown that for maintaining the same Rx power level at a frequency of 28 GHz compared to 900 MHz, we would have to increase the antenna gain by 30 dB. This can be obtained by an antenna array using a high number of antenna elements to obtain a directional characteristic of the antenna array, so called beamforming. More antenna elements (usually M = 2N) results in higher gain and narrow beam width. In addition, beamforming can significantly reduce the energy consumption by targeting individual UEs with their assigned signal. In a normal base station without beamforming, the energy that is not received by the UE gets absorbed into the environment or creates interference for adjacent UEs (see Figure 1).

In existing standards like LTE or WLAN, multiple antenna concepts known as MIMO are used to obtain a higher capacity. This concept of spatial multiplexing exploits the channel decorrelation situation by using multiple antennas at input and output to achieve a higher throughput per user. Multi-user MIMO extends MIMO by sending data to different UEs simultaneously by exploiting their uncorrelated locations through the use of beamforming. The term massive MIMO describes the combination of those concepts, beamforming together with spatial multiplexing in a dynamic manner depending on the circumstances given by hardware and channel conditions (see Figure 2).

 Challenges for Massive MIMO

There are many advantages of using massive MIMO antenna technologies. But there are also several challenges when realizing it. Some of those challenges are:

1) High throughput for fronthaul interface connection:

If a C-RAN architecture is used, the basestation connects to the centralized baseband through a fiber optic network or microwave links. If fiber is used, it is through a digital interface, such as CPRI (Common Protocol Radio Interface) where digital IQ components are transmitted. The required data rate for CIPRI can be calculated as BWCPRI = S•A•ƒS•bS•2•O•LC  with S = number of sectors, A = number of antenna elements (or channels), fS is the A/D sampling ratio, bS is the bits/samples, O is the protocol overhead and LC is the line coding or compression scheme. As seen by this formula, the CIPRI bandwidth requirement scales linearly with antennas and signal bandwidth.

2) Antenna array calibration:

Due to strict requirement of the antenna array beamsteering for precise phase and amplitude differences between the antenna elements, each array must be calibrated for the following tolerances:

Phase: Phase error can have a large effect on the antenna beam depending on its statistical properties. If the phase error is uniformly distributed across the array, then the main beam direction does not change. Instead, the nulls that are often used to block interference are severely affected, losing 10 dB to 20 dB. If there is a more deterministic phase error distribution, then this will steer the beam in a different direction, known as beam squint. Phase error can be caused by manufacturing tolerances in the RF feeding network, thermal effects in the PAs and LNAs, and group delay variations in the filters.

Amplitude: Amplitude error does not affect the direction of the beam, but rather the peak gain and the sidelobe levels, and is generally due to the thermal effects on the active components (PA and LNA).

Timing/Frequency: Depending on the circuit architecture, if a common LO network is not used between modules, there will be frequency drift in addition to the timing errors in the ADCs.

3) Mutual coupling between antenna elements:

An antenna array is formed by placing several antennas adjacent to each other, either in a one-dimensional line or in a two-dimensional plane. An antenna is both a radiator and an absorber of nearby electromagnetic fields. Therefore, adjacent antennas will couple energy to other antennas, PCBs, components, etc. within the reactive near-field region of the antenna. The most common method to reduce the mutual coupling effect is to increase the separation distance between antennas, but these larger separations lead to bigger arrays and higher sidelobes in the antenna far-field (sidelobes increase interterfence for adjacent users).

These coupling effects can be measured with a Vector Network Analyzer (VNA). Important to note is that transmission S-Parameters in between antennas allow for evaluation of the cross coupling. Mutual coupling between antenna elements of an antenna array will decrease the radiated power level and therefore lead to a shorter cell range or the capacity of the system (reduced SNR). The structure of the antenna array contributes directly to the mutual coupling performance. In traditional one-dimensional linear antenna arrays, each single polarized antenna element would have two direct neighbors. For a uniform square array (USA) with two polarizations (e.g. co-pol and cross-pol), however, one antenna element has now up to 17 direct neighbors. Figure 3 shows that the antenna distance between USA elements has to become larger compared to a uniform linear array (ULA) in order to achieve the same channel capacity.

In order to simultaneously measure the coupling between antenna elements in an antenna array, we need a VNA with multiple ports. To increase the number of connectors, one may use a switch matrix that is connected to the DUT. This method, however, requires some switching between the measurements so it may affect the performance and the test duration.

True multiport VNAs are equipped with multiple receivers instead of switches to perform tests simultaneously in order to reduce the test duration and perform a better performing complete mutual coupling measurement between one antenna element and its surrounding neighbors. If the number of antenna elements is higher than the number of simultaneous ports (e.g. higher than the 24 port for the R&S ZNBT8), switch matrixes can also be added. An additional benefit is that specific tests like “active return loss” (S11, S22, S33, …, S2424) can be measured in parallel with many ports stimulated simultaneously. This method provides deeper insights into antenna array in the design phase, which has an effect on the real world operation case in terms of network capacity. Figure 4 shows a simultaneous true multiport measurement performed with the R&S ZNBT20. It clearly shows that the active return loss is significantly different than the passive return loss, thereby requiring the measurement of all adjacent antenna ports simultaneously.

4) Irregular antenna arrays:

The antenna array structure ULA or USA could be replaced in practical implementations by design invisible antenna arrays where the array has the shape of characters, advertising logos, or even religious symbols in order to integrate into the city environment in an aesthetic manner. While these irregular arrays can reduce mutual coupling (fewer adjacent neighbors), there is an impact on the beamforming performance.  In particular, the grating lobes increase and there are more side lobes.  These can be controlled by adjusting the amplitude difference between antenna elements at the cost of increased complexity.

5) Antenna array complexity:

The final challenge discussed in this article is the dramatic increase of complexity compared to the current basestation antenna design. From the beamforming method like analog beamforming, digital beamforming or hybrid beamforming to the direct integration of radio transceivers with antennas, this increases the number of antenna elements, RFICs, Pas, LNAs, filters, switches, and duplexers. These factors will not only determine the cost and implementation complexity but also the RF performance of antenna arrays.

Over the Air (OTA) Measurements

When it comes to testing, there are two major aspects for antenna arrays that explain the necessity for over the air connection: one aspect is the usage of higher frequencies where the simple usage of connectors will no longer be possible due to costs, high losses, and coupling. Another aspect is that massive MIMO systems integrate the radio transceivers directly with the antennas, resulting in a loss of RF test ports. This means that the DUT radio and antenna performance can only be measured with an OTA interface, rather than the traditional conductive interface with a cable. What are the consequences and challenges of this paradigm change? A new measurement dimension has to be introduced: space or direction of departure.

In the past, power was measured as a function of time, spectrum, or code (CDMA systems). With the addition of beamforming, another dimension is added: space or power versus direction of departure. Figure 5 shows these dimensions through an example of a power measurement.

Radiated Fields 

The electromagnetic fields from any antenna can be described and measured in two different regions: near-field and far-field, that are defined by the Fraunhofer distance R = 2*D2/λ. In the near-field region of the antenna, at distances less than R, the field consists of both reactive and radiated components, whereas the far-field of an antenna has only the radiated component. For an OTA system, this means that in order to characterize the antenna radiation performance, the measurement can be performed in either the near-field or the far-field region. In the near-field region, a precise measurement of both the phase and the magnitude of the electromagnetic field over a three dimensional surface surrounding the DUT is required for the mathematical transformation to the far-field region, resulting in the antenna 2D and 3D gain patterns. A measurement in the far-field region only needs the magnitude of the field in order to calculate the beam pattern of the antenna and can be measured at a single point in space if desired. Figure 6 illustrates the electromagnetic fields from a basestation antenna array of eight circular microstrip antenna patches at 2.70 GHz with uniform excitation, where D is the maximum antenna aperture or size.

Over the air measurements can be used to measure both absolute values like transmit radiated power or relative values like receiver sensitivity at a specified SNR. OTA tests measuring the three-dimensional antenna pattern can be performed either in near-field or in far-field. Measurements in near-field allow smaller anechoic chambers for the measurement, but require instruments capable of measuring both phase and amplitude with high location precision and additional post processing for the near-field to far-field transformation.

OTA measurement parameters can be divided into two general categories: R&D for more complete investigation of the DUT radiated properties, and production for calibration, verification, and functional testing as summarized below:

Over the Air Measurements in R&D 

– Gain patterns: Gain patterns are either 2D from one of the three principal planes (E1, E2, or H-plane) or a complete 3D pattern, as in Figure 7.

– Radiated power: The effective radiated power (ERP) or effective isotropic radiated power (EiRP) is used to measure an active antenna system either as a UE or a basestation. For UE testing, total radiated power (TRP) is used instead, where TRP is the weighted integral of the ERP values over a sphere.

– Receiver sensitivity: Receiver sensitivity is characterized by the parameter of effective isotropic sensitivity (EiS) or total isotropic sensitivity (TiS) where EiS can be calculated as Tis plus the antenna directivity. EiS effectively measures the block error rate as a function of the receive power equal to the specified receiver sensitivity.

– Transceiver and receiver characterization: Each individual transceiver in the active antenna system needs to be verified through an OTA interface. This includes a range of measurements for both the transmitter and the receiver as listed in Table 1. It is assumed that each transceiver will turn on for individual verification or a set of transceivers for joint assessment.

– Beamsteering and beam tracking: Due to the high path loss and limited range of a mmWave wireless system, precise beam tracking and fast beam acquisition is required for mobile users. Whereas with antenna implementation for existing cellular technologies, static beam pattern characterization was sufficient, mmWave systems will require dynamic beam measurement systems to accurately characterize beam tracking and beamsteering algorithms.

Over the Air Measurements in Production Testing 

– Antenna/relative calibration: In order to accurately form beams, the phase misalignment between RF signal paths needs to be below ±5° (greater than 5 degrees misalignment typically leads to higher sidelobes, null elimination, and beam misdirection). This measurement can be performed for both passive and active antenna systems using a phase-coherent receiver to measure the relative difference between all antenna elements. This is then compiled into a lookup table or codebook for the AAS to use as a reference for beam generation or to calibrate the internal self-calibration circuits inside the AAS unit.

– Transceiver calibration: Due to the lack of RF ports on some massive MIMO systems, the individual transceivers will also need to be calibrated using OTA techniques, as detailed in Table 1.

– Five point beam test: According to 3GPP, the active antenna system (AAS) manufacturer specifies a beam direction, maximum EiRP, and an EiRP threshold for each declared beam. In addition to the maximum EiRP point, four additional points are measured at the declared threshold boundary (see Figure 8).

– Functional tests: This is the final test performed on the completely assembled unit in production. It can consist of a simple radiated test, a five point beam test, and aggregate transceiver functionality, such as an EVM measurement of all transceivers.

OTA measurement systems can be classified into two distinct types, depending on which part of the radiated field is being sampled, as introduced in Figure 6. The field regions are separated according to the power distribution of the electromagnetic field. In the reactive near-field region, the power is contained within the phase component of the electromagnetic field, whereas the radiated field in the far-field region contains power only in the magnitude of the electromagnetic wave. The region between these two extremes is the radiated near-field, where both the phase and magnitude of the field need to be measured. Measurements are typically not performed in the reactive near-field of the DUT since any object within this area couples to the DUT and becomes part of the effective DUT volume.  Therefore, most measurements are performed either in radiated near-field or the far-field region of the DUT.

For small devices (in terms of wavelengths) such as UEs, the device size is small enough such that the required chamber size for far-field conditions is dominated by the measurement wavelength. For larger devices, such as basestations or massive MIMO, the required chamber size becomes very large. Huygen’s principle in electromagnetics states that if the tangential electric and magnetic fields are known on an arbitrary surface enclosing the antenna, then the equivalent far-field radiation properties can be calculated using Fourier transforms. Chamber sizes can be reduced significantly as long as the measurement system accurately samples the phase and magnitude of the electromagnetic field on the entire enclosing surface.


– Far-field chamber size: Measuring in the far-field region only requires a direct measurement of the magnitude of the plane waves. Such chambers are generally quite large, where the length is set by a combination of the DUT size and the measurement frequencies.

– Near-field chamber size: Although the far-field is generally measured at a suitable distance from the DUT, it is possible to manipulate the electromagnetic fields such that a near-field chamber can be used to directly measure the plane wave magnitudes. There are two possible techniques:

i) Compact range chambers: The simplest method to form a planar wave at the surface of the DUT is to extend the path of the electromagnetic fields by using reflectors, similar to optical reflectors. Due to the expense in constructing accurate reflectors, this technique is used mostly for large DUTs such as aircrafts and satellites.

ii) Plane Wave Converter (PWC): A second method to create a planar wave at the DUT is to replace the measurement antenna with an antenna array. Similar to using lenses in an optics system, the antenna array can generate a planar far-field at a targeted zone in the region of the DUT (see Figure 9).

Radiated Near-Field: 

Measurements in the near-field region require both the field phase and magnitude sampled over an enclosed surface (spherical, linear, or cylindrical) in order to calculate the far-field magnitude using Fourier spectral transforms.

This measurement is usually performed using a vector network analyzer with one port at the DUT and the other port at the measurement antenna. For active antennas or massive MIMO, there are often no dedicated antenna or RF ports, so the OTA measurement system must be able to retrieve the phase in order to complete the transformation into far-field. There are two methods of performing phase-retrieval for active antenna systems (see Figure 10):

– Interferometric: This method uses a second antenna with a known phase used as a reference. The reference signal is mixed with the DUT signal with unknown phase. Using post-processing, the phase of the DUT signal can be extracted and used for the near-field to far-field transformation.

– Multiple surfaces or probes: Instead of using a second antenna for the phase reference, this method uses a second surface volume as the phase reference with at least one wavelength separation between the two measurement radii. As an alternative to the measurement of multiple surfaces, two probes with different antenna field characteristics can be used instead over a single measurement surface. The two probes need to be separated by at least half-wavelength to minimize mutual coupling.


Antenna arrays will play an essential role in future wireless communication systems due to many inherent advantages. But there are some challenges in the development, design, and production process of these massive antenna arrays that may inhibit the expected quality of experience at the end user. Thorough testing is needed to guarantee the proper working of such antenna arrays. Due to the elimination of RF test ports and the use of frequencies in the centimeter and millimeter wave length region, OTA will become an essential tool for characterizing the performance of not just the antenna arrays of an Active Antenna System or a massive MIMO array, but the internal transceivers as well. For this reason, there will be a high demand for OTA chambers and measurement equipment not only to measure the strict radiative properties of antennas, but substituting traditional conducted transceiver measurements as well. Rohde & Schwarz, with its wide range of anechoic chambers and measurement equipment expertise, is well situated to deliver solutions even for future customer requirements.


[1] 3GPP TR 37.842 v2.0.0, “Radio Frequency (RF) Requirement Background for Active Antenna System (AAS) Base Station (BS).”

[2] 3GPP TS 36.141 v13.3.0 TSG RAN E-UTRA; Base Station (BS) conformance testing. – 2016.

[3] 3GPP TS 37.105 v. 13.0.0 TSG RAN; Active Antenna System (AAS) Base Station (BS) transmission and reception. – 2016.

[4] Rowell, Kottkamp Antenna array testing conducted and over the air (1MA286) – 2016

[5] Lloyd, Reil Millimeter-Wave Beamforming: Antenna Arrays and Characterization (1MA276). – 2016


Figure 1: Energy efficiency by using beamforming
Figure 2: Massive MIMO – combination of beamforming and spatial multiplexing
Figure 3: Mutual coupling affecting base station capacity and range
Figure 4: Simultaneous active return loss measurement of 12 antenna elements using an R&S ZNBT20
Figure 5: Power measurements as a function of time, frequency, code, and space
Figure 6: Electromagnetic fields from a basestation antenna array
Figure 7: Example 3D antenna pattern for a small antenna array at 2.1 GHz
Figure 8: Five point test based on manufacturer declaration of 5 measurement points
Figure 9: Plane wave conversion principle