Welcome to the website of Yongshun Technology Co.,Ltd.!Today is:
News Center
Check category

Massive MIMO and beamforming: reveal the signal processing behind 5G

  • Categories:News Center
  • Time of issue:2021-01-27 14:30
  • Views:

(Summary description)Our thirst for high-speed mobile data is endless. However, the available radio frequency (RF) spectrum in dense urban environments has been saturated, and it is clear that there is an urgent need to improve the efficiency of sending and receiving data from wireless base stations.

Massive MIMO and beamforming: reveal the signal processing behind 5G

(Summary description)Our thirst for high-speed mobile data is endless. However, the available radio frequency (RF) spectrum in dense urban environments has been saturated, and it is clear that there is an urgent need to improve the efficiency of sending and receiving data from wireless base stations.

  • Categories:News Center
  • Author:Claire Masterson, ADI Limerick communication system Team System Application Engineer
  • Origin:
  • Time of issue:2021-01-27 14:30
  • Views:

Our thirst for high-speed mobile data is endless. However, the available radio frequency (RF) spectrum in dense urban environments has been saturated, and it is clear that there is an urgent need to improve the efficiency of sending and receiving data from wireless base stations.

The base station contains a large number of antennas. Therefore, one solution to improve the spectral efficiency of the base station is to allow these antennas to communicate with multiple spatially separated user terminals simultaneously through the same frequency resource, and use multipath transmission. This technique is often referred to as massive MIMO (massive MIMO). You may have heard that massive MIMO is described as beamforming of a large number of antennas. But the question that follows is: What is beamforming?

The relationship between beamforming and Massive MIMO

Flexible application of optocouplers in green energy and energy storage systems

Different people have different understandings of the term beamforming. Beamforming refers to the ability to automatically adjust the radiation pattern of an antenna array according to a specific scenario. In the field of cellular communication, many people think that beamforming is to direct the main lobe of antenna power to the user, as shown in Figure 1. Adjust the amplitude and phase of each antenna transceiver unit so that the transmit/receive signals of the antenna array in a specific direction are uniformly superimposed, while the signals in other directions cancel each other out. Generally, the spatial environment of the array and the user is not considered. This is indeed beamforming, but only a specific implementation of it.

Figure 1: Traditional beamforming

Massive MIMO can be regarded as a form of beamforming in a broader sense, but it is far from the traditional form. Massive means a large number of antennas in the base station antenna array; MIMO means that the antenna array uses the same time and frequency resources to meet the needs of multiple users separated in space. Massive MIMO also believes that in the actual system, the data transmitted between the antenna and the user terminal (and the reverse process) is filtered by the surrounding environment. The signal may be reflected by buildings and other obstacles. These reflections may involve delay, attenuation, and direction of arrival, as shown in Figure 2. There may not even be a direct path between the antenna and the user terminal. As a result, these indirect transmission paths are also valuable.

Figure 2: Multipath environment between antenna array and user

In order to utilize multipath, the spatial channel between the antenna element and the user terminal needs to be characterized. The literature generally refers to this response as channel status information (CSI). This CSI is essentially a set of spatial transfer functions between each antenna and each user terminal. Use a matrix (H) to collect this spatial information, as shown in Figure 3. The next section will discuss the CSI concept and its collection method in detail. CSI is used to digitally encode and decode the data sent and received by the antenna array.

Figure 3: Characterization of a massive MIMO system requires channel status information

Characterize the spatial channel between the base station and the user

Imagine an interesting analogy: a balloon is punctured in a certain position, making a "pop" sound, and recording this sound or pulse in another position, as shown in Figure 4. The sound recorded at the microphone position is a spatial impulse response, and the information contained therein is unique to the specific location of the balloon and the microphone in the surrounding environment. Compared with the direct path, the sound reflected by obstacles will have attenuation and delay.

Figure 4: Illustrate the spatial characteristics of the channel through a sound metaphor

If the analogy is extended to compare the antenna array and the user terminal scenario, then more balloons are needed, as shown in Figure 5. It is worth noting that in order to characterize the channel between each balloon and the microphone, we must burst each balloon at different timestamps so that the balloon reflections recorded by the microphone do not overlap. The other direction also needs to be characterized, as shown in Figure 6. In this example, when the balloon at the user terminal is punctured, all recordings can be completed at the same time. Obviously, it takes much less time!

Figure 5: Representation of a downlink channel with a sound metaphor

Figure 6: Representation of uplink channel with sound metaphor

In the RF field, pilot signals are used to characterize spatial channels. The OTA transmission channel between the antenna and the user terminal is reciprocal, that is, the channel is the same in both directions. It depends on whether the system is operating in time division multiplexing (TDD) mode or frequency division multiplexing (FDD) mode. In TDD mode, uplink and downlink transmissions use the same frequency resources. The assumption of reciprocity means that only the channel needs to be characterized in one direction. The uplink channel is the obvious choice, because only one pilot signal needs to be sent from the user terminal, which can be received by all antenna elements. The complexity of channel estimation is proportional to the number of user terminals, not the number of antennas in the array. This is very important because the user terminal may be on the move, so channel estimation must be performed frequently. Another important advantage of uplink-based characterization is that all the heavy channel estimation and signal processing tasks are completed at the base station, not at the user end.

Figure 7: Each user terminal transmits orthogonal pilot symbols

Now that the concept of collecting CSI has been established, how can this information be applied to data signals to support spatial multiplexing? Filtering is designed based on CSI to pre-code the data transmitted by the antenna array, so that the multipath signals will be uniformly superimposed at the user terminal position. This filtering can also be used to linearly combine the data received on the RF path of the antenna array to detect data streams from different users. This issue will be discussed in more detail below.

Supports Massive MIMO signal processing

The previous section describes how to estimate CSI (represented by matrix H). The detection and precoding matrix is ​​based on H calculation. There are multiple calculation methods for this matrix. The following will focus on the linear solution. Examples of linear precoding/detection methods are maximum ratio (MR), return to zero (ZF), and minimum mean square error (MMSE). At the end of this article, we will provide the whole process of exporting precoding/detection filters from CSI, and discuss its optimization standards and the advantages and disadvantages of each method.

For the above three linear methods, Figure 8 and Figure 9 show the operation of signal processing in the uplink and downlink respectively. For precoding, there may be some kind of scaling matrix to meet the power standard that the entire array is ignored due to simplification.

Figure 8: Uplink signal processing; H represents conjugate transpose

Figure 9: Downlink signal processing; T means transpose; means conjugate

As the name implies, maximum ratio filtering is designed to maximize the signal-to-noise ratio (SNR). From the perspective of signal processing, this is the simplest method, because the detection/precoding matrix is ​​just the conjugate transpose or transpose of the CSI matrix H. Its biggest disadvantage is that it ignores interference between users.

The return-to-zero precoding attempts to solve the problem of interference between users and minimize it by designing optimization standards. The detection/precoding matrix is ​​the pseudoinverse of the CSI matrix. The computational cost of the pseudo-inverse matrix is ​​higher than the complex conjugate in the MR case. However, due to too much emphasis on reducing interference, the user's received power will be affected.

MMSE tries to strike a balance between amplifying the signal and reducing interference. The price to pay for this holistic view is the complexity of signal processing. The MMSE approach introduces a regularization project for optimization-denoted as β in Figures 8 and 9-using it to find a balance between noise covariance and transmit power. This method is sometimes called Normalized Return to Zero (RZF) in the literature.

The above does not cover all precoding/detection techniques, but simply introduces the main linear methods. In addition, non-linear signal processing techniques, such as dirty paper encoding and continuous interference cancellation, can be used to solve this problem. These methods provide optimal capacity, but they are very complicated to implement. The above linear method is generally sufficient for massive MIMO, and the number of antennas can be large. The choice of precoding/detection technology depends on the variety of computing resources, the number of antennas, the number of users, and the environment in which the system is located. For large antenna arrays where the number of antennas is much larger than the number of users, the maximum ratio method may be sufficient to meet the needs.

Real world system challenges Massive MIMO

There are other practical issues to consider when implementing massive MIMO in real-world scenarios. For example, an antenna array has 32 transmit (Tx) channels and 32 receive (Rx) channels, operating in the 3.5GHz frequency band, then 64 RF signal chains need to be placed. At the specified operating frequency, the antenna spacing is about 4.2 cm. This means that there is a lot of hardware that must be packed into a small space. It also means that a lot of power will be dissipated, which will inevitably cause temperature problems.
Figure 10 shows the downlink channel in the real world system. It is divided into three parts: Over-the-air download (OTA) channel (H), the hardware response (T BS ) of the base station transmitting the RF path, and the hardware response (R UE ) of the user receiving the RF path . On the uplink, in contrast, R BS characterizes the base station receiving hardware RF path, and T UE characterizes the user transmitting hardware RF path. Although the assumption of reciprocity holds for the OTA interface, it does not hold for the hardware path. The RF signal chain can cause errors in the system due to problems such as mismatched wiring, poor synchronization between RF paths, and temperature-related phase drift.

Figure 10: Real-world downlink channel

All local oscillator (LO) phase-locked loops (PLLs) in the RF path use a common synchronous reference clock, and use synchronous SYSREF for the fundamental digital JESD204B signal to help solve the delay problem between RF paths. However, when the system is started, there is still a phase mismatch between channels between the RF paths, and the phase drift caused by temperature will further amplify this problem. Therefore, the system obviously needs to initialize the calibration at startup, and perform periodic calibration in the subsequent operations. The advantage of reciprocity can be achieved through calibration, so that the signal processing complexity is maintained at the base station, and only the uplink channel needs to be characterized. This can be simplified in a general sense, so that only the base station RF paths (T BS and R BS ) need to be considered .

There are many ways to calibrate these systems. One is to place a calibration antenna in front of the antenna array, and use this calibration antenna to calibrate the receiving and transmitting RF channels. However, whether this method of placing antennas in front of the array meets the requirements of actual system calibration is still questionable. Another method is to use the interactive coupling between existing antennas in the array as a calibration mechanism, which is highly feasible. Perhaps the simplest and most direct method is to add some passive coupling paths before the antenna in the base station. This will increase the complexity of the hardware, but it should provide a more durable calibration mechanism. In order to fully calibrate the system, when a signal is sent from a designated calibration transmitter channel, it will be received by all RF receiving paths connected through passive coupling. Then, each transmit RF path sends a signal in sequence, and is received at the passive coupling point in front of each antenna, and then transmitted back to the combiner, and then sent to the designated calibration receiving path. Temperature-related effects generally change slowly, so unlike channel characteristics, there is no need to perform temperature-related calibration frequently.

For example, ADI's integrated transceivers provide high-efficiency solutions to such problems. This series of products is particularly suitable for applications that require high-density RF signal chains. For example, AD9371 integrates 2 transmission paths, 2 reception paths and 1 observation receiver in a 12mm×12mm package, as well as 3 fractional-N PLLs. To generate RF LO. The high level of integration allows manufacturers to create complex systems in a timely and cost-effective manner.

Figure 11 shows a possible system configuration using multiple AD9371 transceivers. The system is equipped with 16 AD9371 transceivers, providing 32 transmitting channels and 32 receiving channels. Three AD9528 clock generators provide the PLL reference clock and JESD204B SYSREF for the system. AD9528 is a dual-stage PLL that provides 14 LVDS/HSTL outputs and integrates a JESD204B SYSREF generator, which can be used for multi-component synchronization. The AD9528 adopts a fan-out buffer configuration, one of which is used as the master element, and some other outputs are used to drive the clock input and the SYSREF input of the slave element. Including a possible passive calibration mechanism (shown in the green and orange parts in the figure), a dedicated transmit and receive channel through the splitter/combiner to calibrate all receive and transmit signal paths.

Figure 11: 32T/Rx massive MIMO RF front end with ADI AD9371 transceiver


Massive MIMO spatial multiplexing is expected to become a revolutionary technology that rewrites the rules of the game in the field of cellular communications, supporting the realization of higher mobile capacity and efficiency in high-traffic urban areas. It takes advantage of the diversity brought about by multipath propagation, allowing data transmission between the base station and multiple users at the same time and frequency resources. The channel between the base station antenna and the user is reciprocal. Therefore, all complex signal processing can be kept at the base station, and the channel characterization can also be done in the uplink. For example, the ADI RadioVerse series of integrated transceiver products support the realization of high-density RF paths in a small space, and are therefore very suitable for massive MIMO applications.


Scan the QR code to read on your phone

Follow Us:

Home         Message

About         Contact

News         ENGLISH

Product      Japanese


Follow us
qr Chinese
qr English
qr Japanese

Contact us

Shenzhen Xinxie Office
Contact number: 18998036969
(Yang Sheng)
Contact email: benson@szxinxie.cn
Taiwan Jie Shun Office
Contact number: 09-8166-0710
(General Manager-Yang Fengyuan)
Contact email: kent@yong-shun.com.tw

Company website: https://www.rfxinxie.com Company Address: No. 5, Alley 18, Lane 1, Renxing Road, Hukou Township, Hsinchu County

 Copyright © 2019 Xinxie Technology (Shenzhen) Co., Ltd.  粤ICP备20005589号

Username used for comment: