Detailed explanation of 5G OTA testing technology

btty038

Detailed explanation of 5G OTA testing technology [Copy link]

1. What is OTA (Over The Air)

OTA stands for Over The Air. In order to test a device using any test equipment, you need a way to connect the device to the test equipment. OTA is a method of connecting a device to the test equipment. There are roughly two connection methods as shown below. One is Conductive and the other is Radiative (or OTA). In short, OTA is a connection method through a pair of antennas (transmitting antenna and receiving antenna).

Conductive	Radiation/OTA

OTA is actually a very complex topic. There are many different aspects that can be thought of. I will try to cover as many different perspectives as possible as I learn and experience more.

2. Types of OTA settings

When we say radiated testing, it generally refers to a variety of different types of configurations, as shown below. These are just some of the typical examples you'll see most often, but they're not all. There are a lot of different variations of radiated test setups. Although the terms OTA testing and radiated testing are used interchangeably, when we say OTA testing without any specific details, we're generally thinking of a configuration like (C) or (D) shown below. As shown in the image below, it's tested in a chamber lined with absorbers (this type of chamber is called an anechoic chamber. Anechoic means 'no echoes.' 'Anechoic' in this case means 'no reflections from anything in the box').

(A)	(B)
This is probably one of the simplest methods. We use a wide flat patch antenna and place the DUT on top of the antenna pad. Very convenient for tests like protocol or functional testing that don't require precise RF measurements. However, unless you don't use this setup inside a shielded room, it may be subject to interference from the surrounding environment (e.g. from a live network or other devices)	In terms of antenna setup, this is almost the same as (A). But in this case, the antenna pad and UE are located in a small RF room. The benefit of this type compared to (A) is that it can block interfering signals (for example, interference from a live network or neighboring devices)

(C)	(D)
This is a chamber made of a conductive metal (usually copper). This blocks interference from the surrounding environment, as in (B), while it reduces interference from reflections from the sides of the box. When electromagnetic waves from the DUT or device hit the conductive walls, the conductors block most of the waves.	This would be an ideal solution for OTA testing where a certain level of measurement accuracy is required. It is lined with special absorbers and usually has positioners that can change the orientation of the DUT via an external controller.

3. Why choose OTA?

Even in legacy technologies (e.g., UMTS, LTE), we sometimes do OTA measurements, especially for TRP or TIS measurements. However, in 5G/NR, we are talking about OTA for almost every test, even for protocol tests. Why has OTA become such a big issue in 5G/NR? In NR, there are about two separated spectrums specified in the 3GPP specifications. One is FR1 (sub 6 Ghz) and the other is FR2 (mmWave). In FR1, we can continue to do conductive testing, just like with 2G/3G/4G technologies. However, in FR2, most likely we are forced to work with OTA. Why?

We can think of several reasons and some different aspects.

• Complexity : In FR2, we will almost certainly be using some type of array antenna (called Massive MIMO). This means that you will have a large number of antennas on the device. If you want to do a conductive test, the connections will be as shown below (B), while if you choose OTA, you can test like in (A). Then, it is obvious why we chose (A). Note: If you are wondering why we need to use antenna arrays, the motivation for Massive MIMO page will give you some insights.
• Not enough space : Let’s assume you have a good reason to go OTA, despite the complexity of a wired connection, you’ll still face another serious problem. Even if you have many antenna elements in your antenna array (e.g., 16, 32, 64, etc.), the overall size of the antenna module is not large enough at mmWave frequencies to accommodate all the cable connectors.
• Cost : Now let's assume that you have a very, very serious reason to do conductive (say B), despite the complexity and space issues. Even in this case, there are other problems with conductive testing. In most conventional testing, you may have used low-cost SMA connectors and cables. However, you cannot get accurate measurements using SMA type connectors/cables in mmWave. You need K connectors or more special connectors and cables (e.g., V connectors) if the frequencies get higher. The cost of these types of special connectors and cables is much higher than those SMA types. If we need to use very high frequencies in the future (e.g., over 60 Ghz), you may need to spend almost as much money on connectors and cables, rather than the low-cost equipment price.
• The physics of the measurements : Even if you overcome all of the above issues, some types of measurements require OTA due to the nature of the measurements themselves. For example, if you want to detect the direction of the beam formed by an antenna array, you must rely on OTA measurements. You could argue that you could still do this with conductive testing. In theory, you could downlink all the signals from each antenna element path to baseband and figure out the beam direction (and other properties of the beam) through baseband processing. Sure, in theory this is possible. But I'm 100% sure you want to avoid doing that if there is a relatively simple way to do it like OTA testing.

(A)

(B)

4. UE placement in the test setup (antenna distance between UE and test equipment)

In order to obtain repeatable, reliable and stable measurement results, it is very important to place the AUT (antenna under test) and the measurement antenna in the appropriate position. In this section, I will explain how to determine the appropriate antenna position and the theoretical background why a specific position should be used.

The minimum far-field distance R of a traditional far-field anechoic chamber is determined by the following formula a (based on R5-180013).

< Figure 1: Antenna field area >

The near/far field boundaries for different antenna sizes and frequencies are shown in the table below. This table is based on R5-180013 (reference [1]) - Table 2.2.1: Near/far field boundaries for different frequencies and antenna sizes in a traditional far-field anechoic chamber

< Table 1 - Near/far boundary distances of D and frequency >

D (cm)	Frequency (GHz)		Far/near boundary (cm)	Path loss
5	28	47	48	55
10	28	187	188	66.9
15	28	420	420	73.8
20	28	747	748	78.9
25	28	1167	1168	82.7
30	28	1680	1680	85.9
5	100	167	168	76.9
10	100	667	668	88.9
15	100	1500	1500	96
20	100	2667	2668	101
25	100	4167	4168	104.8
30	100	6000	6000	108

NOTE : From this is optional reading. I have dug deeper into this equation out of curiosity. If you are not interested you can skip this part. I was trying to investigate how the near/far boundaries change with frequency.

As shown below, the span of the radiated near field (the gap between the reactive near field and the start of the far field) increases dramatically with increasing frequency, where the reactive near field distance increases relatively slowly). Also, comparing the following two graphs, you will notice that as D increases, the far field distance becomes very large.

< Figure 2 - Field boundary variation with frequency at D = 5cm >

< Figure 3 - Field boundary variation with frequency at D = 10cm >

Now you may have an interesting question. According to the figure above, the distance between the DUT antenna and the device antenna should increase as the frequency increases. That is, as the frequency increases, the size of the anechoic chamber will increase? Doesn't this make intuitive sense to you? Our common sense (our RF intuition) says that the size of frequency-dependent objects tends to decrease as the frequency increases.

How do we deal with this conflict using our intuition and the plot above?

The solution is that D is not a constant. In the case shown in the figure above, D has a fixed value regardless of the frequency. But when we design antennas, we usually determine the value of D (antenna aperture size) in terms of wavelength, as shown below. Here, k is just a constant, such as 0.5, 1, 2, etc.

If you plot the wavelength (lamda) as the frequency increases, you will get a graph as shown below. You will notice that the wavelength decreases to a great extent.

< Figure 4 - Wavelength and Frequency >

If you rewrite the far field distance equation, it would look like this. In this equation, you'll notice that the far field distance decreases as the frequency increases. (Note: If you want to try to calculate in real terms, think of f as 'frequency in Hz', c as 'speed of light in m', and k is just a constant without any units).

5. Why test in the far field?

Probably by now you might have a question ‘why do we need to test in the far field?’. It is not easy to answer this question directly. So let me change the question a little bit. Why don’t we test in the near field?

The simple answer to this question is that measurements in this area tend to be unpredictable and vary with small changes in the antenna's surroundings (e.g., the circuits feeding the antenna) and changes in position. In contrast, the field pattern in the far field is more stable and predictable, and is less sensitive to small surrounding components.

For those interested in further details, let's take another look at the characteristics of each region. You can investigate further on your own. Try Googling keywords like "near field and far field", "field region around antenna", etc.

Reactive Near Field : This is the region very close to the antenna. The relationship between the E and H fields in this region is very unpredictable (it is unpredictable not because this property defies physical theory, but because the physics is so complex). For example, at one point you will see the E field dominate, and at another point the H field dominates except at the previous point. The radiated energy also goes back and forth and affects the surrounding electrical components just like the antenna control circuit. For example, some portion of the radiated energy is absorbed and stored in the surrounding components at some point in time, and the stored energy is radiated back into space at other points in time and affects the radiation pattern.

Radiating Near Field (Fresnel Zone) : In this region, the distance to the antenna is not too close to the influence of anti-electronic components as mentioned above, and the E and H field relationship is more predictable compared to the reactive near field. However, the E and H field relationship is still quite complex, and there is a high probability that some physical object may affect the radiation pattern in this region. For example, some metallic object such as a steel beam supporting the antenna module can act as a kind of antenna or reflector. Therefore, such an object can affect the radiation pattern of the AUT (antenna under test).

FAR field ( Fraunhofer region) : In this region, the angular field distribution is essentially independent of the distance from the antenna, and the radiation pattern can be approximated by a spherical wavefront. Since any receiving point in this region is very far from the antenna, the size and shape of the transmitter is no longer important and it can be approximated as a point source. The electric and magnetic fields are in phase, perpendicular to each other and also perpendicular to the propagation direction. In this region, you can safely assume that the wavefront passing through the receiving antenna is planar (i.e., all incident rays are parallel to each other). In simple terms, this is an ideal region where most measurements can be done easily and reliably.

6. Why is antenna size so important?

As mentioned above, in order to obtain stable measurement results, it is important to make the distance between the DUT antenna and the device antenna larger than the far-field boundary. As shown in [Figure 1], the far-field boundary starts at the following distance.

From this expression, you can see that the distance is proportional to D squared (D represents the antenna size). That is, even a small change in D will cause a huge change in distance. To give you a more intuitive understanding, I have plotted this equation in the graph shown on the left. The path loss at the boundary also increases as D increases, as shown in the graph on the right.

< Figure 5 - D > Far field distance and path loss

If you want accurate quantitative data, I put the table as shown below. From this table the two graphs shown above were drawn.

< Table 2 -D's impact on far-field distance and path loss >

D (cm)	Frequency (GHz)		Path loss
1	28	2	27.42
2	28	7	38.30
3	28	17	46.01
4	28	30	50.94
5	28	47	54.84
6	28	67	57.92
7	28	91	60.58
8	28	119	62.91
9	28	151	64.98
10	28	187	66.83
11	28	226	68.48
12	28	269	69.99
13	28	315	71.36
14	28	366	72.67
15	28	420	73.86
16	28	478	74.99
17	28	539	76.03
18	28	605	77.03
19	28	674	77.97
20	28	747	78.86
twenty one	28	823	79.70
twenty two	28	903	80.51
twenty three	28	987	81.28
twenty four	28	1075	82.02
25	28	1167	82.74
26	28	1262	83.42
27	28	1361	84.07
28	28	1463	84.70
29	28	1570	85.31
30	28	1680	85.90

This means that you need to know the exact antenna size to get accurate measurements. However, it is not always easy to correctly define the antenna size. The antenna size D is defined as the maximum distance of the entire antenna module. The red arrow in [Figure 6] indicates D. As you can see, defining D in the cases of (A), (B), (C), and (E) would be straightforward. However, in the case of (D), defining the size is not as easy. In (D), the physical size is the same as (A), but you can see that some conductive material around the antenna module affects the radiation pattern of the antenna module. This can affect the effective size of the antenna, and it is difficult to accurately estimate the effective size. In addition, as shown in (F), (G), and (H), UE manufacturers will allocate antenna modules in several different locations inside the UE. Of course, the shape and location of the antenna module inside the UE will be more diverse and complex than what is shown here.

< Figure 6 - Antenna dimensions for various configurations >

There is another reason why defining D becomes difficult. It becomes even more difficult in the UE case. In order to correctly define D, you need to have all the details of the antenna structure and location in the UE. But in many cases, this information is considered highly confidential by most UE manufacturers. Therefore, when you get a UE (especially a commercialized UE), it is almost impossible to get a precise estimate of D (dimension).

Now we are faced with a very tricky situation. How can we guarantee accurate measurements when we don't have detailed information about the antenna dimensions?

This is what I will discuss in the next section.

7. Handling known D situations and unknown D situations (Whitebox and Blackbox methods)

Before discussing specific cases, let’s think about the meaning of a few basic words - whitebox and blackbox. A white box refers to a box where we can directly see the box and know clearly what is inside the box. This means that the UX knows all the information about the box. A black box refers to a box where we cannot directly see the inside of the box. The only way we can find out the contents of the box is through indirect observation, such as shaking the box, listening to the sound, or lifting and estimating the weight. Of course, this is not the formal definition of whitebox and blackbox in OTA. , but the basic idea applies to the formal definition described in R4-1708553 (reference [4]).

For the “ black box ” approach, the exact antenna location/Centre of Radiation Reference Point (CORRP) does not need to be known.

• • UE positioning has common reference points similar to existing SISO OTA test cases

  • The execution of test cases is of relatively low complexity (no relocation to CORRP is required)

  • The MU element of "offsetting the DUT phase center from the QZ (quiet zone) center" needs to be added for the DUT phase of the MU budget, which depends on the size of the QZ and the range length

For a " white box " approach, the exact antenna location/centre of radiation reference point (CORRP) may need to be known, perhaps through a manufacturer declaration.

• • The CORRP of the active antenna array needs to be aligned with the center of the quiet zone, which can lead to complex execution of the test case

  • It is not necessary to add MU element for "Offset DUT Phase Center from QZ Center (Quiet Zone)" for DUT level, but it is necessary to add MU element for UE relocation

Now you might ask ‘why are we talking about whitebox/blackbox concept here? ’ and ‘how do they deal with D (antenna size)? ’. Let’s think about the case where we have detailed information about D and the case where we don’t have exact information about D.

We assume that we have all the detailed information about D. This means that the antenna module under test is a kind of white box. In this case, we can calculate the exact location of the near/far boundaries. We can then get relatively accurate measurements with minimal distance between the DUT and the measurement antenna (probe) and still meet the far-field criteria in [Figure 6] (A). This means that we can meet the far-field conditions with minimal anechoic chamber dimensions. In turn, this means that we can get accurate measurements with minimal cost on the anechoic chamber.

Now let's consider the case where we don't have accurate information about D. How can we estimate the exact location of the near/far boundary (i.e. the start of the far field)? The answer is 'there is no way to do it'. So how can we guarantee that the antenna is in the far field? The simplest way is to place the receiving antenna far away from the DUT so that you can assume it is in the far field regardless of the size of the antenna as shown in Figure 6(B). Of course there should be a certain limit on the size of the antenna you assume. You would not assume the antenna size is 20 cm when your phone size is 10 cm.

<Figure 6: Possible antenna positions in white-box and black-box conditions>

Summarize,

If we have access to detailed information about the antenna size and exact antenna position on the DUT (e.g., a mobile phone), then using the Whitebox approach is better because we can perform the measurements with a smaller chamber at a relatively low cost. This can be a good option during the development phase, where this information is often open.

If detailed information about the antenna size and the location of the phone is available, the Blackbox method will be a better choice. Since most mobile phone manufacturers are very reluctant to open the detailed information of the antenna on their commercialized devices, the Blackbox method may be the only choice for commercialized devices. However, as mentioned above, we need a very large chamber to apply the Blackbox method, which leads to cost and space issues. To alleviate this problem, another concept is proposed and this alternative is explained in the next section.

Note: Regarding the use of the Whitebox method or the Blackbox method, Ref [3] states the following:

For conformance testing, 3GPP has decided that only black box methods can be used. This is because UE vendors who do not wish to declare antenna structures do not accept the requirements of white box testing.

8. Simulate a not-too-big black box - CATR

As mentioned above, since UE manufacturers do not like to disclose detailed antenna information, only the black box method may be accepted as a test method for commercialized devices, but the black box method tends to require a huge chamber (i.e., a large distance between the transmitter and the receiving antenna). In order to reduce the problem of chamber size, another concept called CATR (Compact Area Test Range) was issued. The overall concept is described in TR 37.842 as shown below.

<TR 37.842 - Figure 10.3.1.1.3.1-1: CATR measurement system setup for EIRP>

As shown in the figure above, you will see that the signal from the transmitter antenna is reflected (reflected) from the specially designed reflector and then reaches the receiver antenna. This will have the effect of folding a long linear distance into a small space, which leads to reducing the size of the chamber. In addition, by designing the reflector in a specific form, you can make all the parallel rays from the DUT reach the feed antenna (measurement antenna). And, you can also make the light from the feed antenna reach the DUT as parallel rays. Actually, the basic principle is similar to the parabolic mirror ray diagram you learned in high school physics. Try googleing 'parabolic mirror ray diagram' or 'parabolic mirror ray tracing' etc.

9. SS-MPAC (Simplified Sector Multi-probe Anechoic Chamber)

The concept of SS MPAC is to use multiple probes around the UE, as shown below, to simulate more realistic radio channel conditions.

<R4-1706669：Figure 2. Detectors and UE installed in a sector>

The main motivations for SS MPAC are described in detail in R4-1706669 as follows:

1. Real-time system performance evaluation, instant communication

2. Ability to simulate real radio channels, meaning the real angular distribution of waves, whether irradiating EU or being irradiated by it

3. Uplink and downlink performance, or reception and response in a multi-node configuration

10. Quiet Area

According to reference [9], the quiet zone is the volume in any chamber where the DUT is illuminated with nearly uniform amplitude and phase. Typical quiet zone specifications are 10 degrees of phase variation, ±0.5 dB amplitude ripple, and 1 dB amplitude taper, which is the roll-off toward the edge of the quiet zone.

Original English link: http://www.sharetechnote.com/html/5G/5G_OTA.html

Reference

[1] 3GPP TSG-RAN WG5 Adhoc Meeting#1 - R5-180013 : Signaling NR Testcases - OTA chamber requirements

[2] Near and far field (Wikipedia)

[3] Keysight Technologies - OTA Test for Millimeter-Wave 5G NR Devices and Systems (White Paper)

[4] 3GPP TSG-RAN WG4 Meeting #84 - R4-1708553: Far field definition and proposal for alternate RF baseline with deterministic antenna array positioning

[5] 3GPP TSG RAN WG4 Meeting NR#2 - R4-1706617: Center of Radiation Reference Point – Reference Definition for OTA Measurements of Phased Array Beamforming Patterns

[6] 3GPP TR 37.842 V13.2.0 (2017-03) - Radio Frequency (RF) requirement background for Active Antenna System (AAS) Base Station (BS) (Release 13)

[7] 3GPP TSG-RAN WG4 Meeting NR AH#2 - R4-1706669: SS MPAC for RRM/Demod

[8] TR 37.977 - Verification of radiated multi-antenna reception performance of User Equipment (UE)

[9] OTA Test for Millimeter-Wave 5G NR Devices and Systems (Keysight Whitepaper)