5G beamforming: an engineer’s overview

5G delivers a step change in network performance over current 4G levels, with peak data rates up to 20 times faster at 20 GB/s, and connection densities of 1000 devices per square kilometre, 100 times more than 4G. This performance improvement is delivered by 5G New Radio (NR), which uses a number of advanced techniques including mmWave (between 30 and 300 GHz), frequency transmissions, advanced signal coding techniques (OFDM), multi-access edge computing (MEC), and network slicing.

Two technologies in particular, Massive MIMO and beamforming, are fundamental to 5G’s enhanced throughput and capacity and work so closely together that they are often described interchangeably. Both are, in fact, complex techniques, meriting separate descriptions. This article concentrates on beamforming, and you can read about Massive MIMO here.

Figure 1: Massive MIMO and beamforming (Source: Ericsson)

Beamforming and Massive MIMO

Beamforming and MU MIMO work together to deliver 5G’s demanding throughput and connection densities (see Figure 1, right).

Massive MIMO (multiple input multiple output) uses multi-antenna arrays and spatial multiplexing to transmit independent and separately encoded data signals, known as "streams". These enable simultaneous communications with multiple user equipment (UE) over the same time period and frequency resource.

Beamforming is used in tandem with MIMO to focus the beams more tightly towards individual UE, enabling higher connection densities and minimising interference between individual beams.

Figure 2: Phased array antenna systems enable beamforming and steering (Source: Analog Devices)

Beamforming Overview

Beamforming is used with phased array antennae systems to focus the wireless signal in a chosen direction, normally towards a specific receiving device. This results in an improved signal at the user equipment (UE), and also less interference between the signals of individual UE.

Phased antenna arrays are designed so that the radiation patterns from each individual element combine constructively, with those from neighbouring elements forming an effective radiation pattern - the main lobe - which transmits energy in the desired direction.

At the same time, the antenna array is designed so that signals sent in undesired directions destructively interfere with each other, forming nulls and side lobes.

The overall antenna array system is designed to maximize the energy radiated in the main lobe, whilst limiting the energy in the side lobes to an acceptable level.

The direction of the main lobe, or beam, is controlled by manipulating the radio signals applied to each of the individual antenna elements in the array.

Each antenna is fed with the same transmitted signal but the phase and amplitude of the signal fed to each element is adjusted, steering the beam in the desired direction (see Figure 2, left).

Fast steering of the beam is achievable since the phase and amplitude of each signal are controlled electronically, allowing adjustments to be made in nanoseconds.

Analogue, digital and hybrid beamforming

There are three methods of implementing antenna beamforming:

Analogue beamforming (see Figure 3, below) is the simplest method, with the signal phase being changed in the analogue domain. The output from a single RF transceiver is split into a number of paths, corresponding to the number of antenna elements in the array. Each signal path then passes through a phase shifter and is amplified before reaching the antenna element.

Figure 3: Analogue beamforming (Source: Commscope)

This is the most cost-effective way of implementing beamforming, since it uses a minimal amount of hardware, however an analogue beamforming system can only handle one data stream and generate one signal beam, limiting its effectiveness in 5G, where multiple beams are required.

In digital beamforming, each antenna element is fed by its own transceiver and data converters (see Figure 4, below), and each signal is pre-coded (with amplitude and phase modifications) in baseband processing before RF transmission.

Figure 4: Digital beamforming (Source: Commscope)

Digital beamforming enables several sets of signals to be generated and superimposed onto the antenna array elements, enabling a single antenna array to serve multiple beams, and hence multiple users. Although this flexibility is ideal for 5G networks, digital beamforming requires more hardware and signal processing, leading to increased power consumption, particularly at mmWave frequencies, where several hundred antenna elements are possible.

Hybrid beamforming (see Figure 5, below) - where analogue beamforming is carried out in the RF stage, and digital beamforming in the baseband - offers a compromise between the flexibility of digital beamforming and the lower cost and power consumption of analogue.

Figure 5: Hybrid beamforming (Source: Commscope)

Hybrid beamforming is recognised as a cost-effective solution for large-scale, mmWave antenna arrays and various architectures are being developed for gNB, (5G base station) implementations. These architectures divide broadly into fully connected, where each RF chain is connected to all antennas; and sub-connected or partially connected, in which each RF chain is connected to a set of antenna elements. Each architecture aims to reduce the hardware and signal processing complexity, while providing near optimal performance: the closest to that of pure digital beamforming.

In all architectures, communication between the gNB and the UE is coordinated using a technique known as beam sweeping, along with Synchronisation Signals (SS), and Channel State Information (CSI), obtained via a Channel State Information Reference Signal (CSI-RS) - a type of pilot signal sent from the gNB to the UE.

Figure 6: 5G networks are moving to a Centralized RAN structure (Source: ISE Mag)

In beam sweeping, the gNB transmits bursts at regular intervals in different spatial directions. The UE listens for these bursts and uses the CSI to determine a channel quality associated with each one.

This quality information is used by the UE to select the optimal beam from its point of view, and the UE informs the gNB of this choice. The UE and the gNB exchange other information - such as analogue or digital beamforming capabilities, beamforming type, timing information - and configuration information - adding to the overhead on the channel.

Hybrid beamforming, with its partitioning of digital and analogue beamforming aligns well with ongoing developments to disaggregate and virtualise the RAN. Centralized RAN, (C-RAN), splits the base stations into low power and low complexity remote radio heads (RRHs) coordinated by a central unit (CU) located at the central office (CO) (Figure 6).

Sharing baseband resources across multiple RRHs makes C-RAN architectures both cost-effective and energy-efficient, making them an attractive option for network densification.

Additionally, removal of the baseband functionality facilitates the deployment of smaller RRHs, which can be flexibly deployed at locations such as lamp posts, electricity pylons and building corners = again supporting network densification and so on.
 

Benefits of beamforming

Beamforming effectively uses the science of electromagnetic interference to enhance the precision of 5G connections, working in tandem with MIMO to improve throughput and connection density of 5G network cells.

The resultant highly directional transmissions are particularly beneficial with mmWave transmissions, which suffer heavily from path loss and do not propagate well through obstacles such as walls. The improved Signal-to-Noise Ratios (SNR), enabled by beamforming, increase signal range for both outdoors and - importantly - indoor coverage.

Beamforming’s ability to cancel out or “null” interference is also a significant benefit in crowded, urban environments with high densities of UEs, where multiple signal beams can potentially interfere with each other.

Overall, by reducing internal and external interference and reducing SNR, beamforming supports higher-order signal modulation schemes, such as 64QAM and 16QAM - all of which contribute to a substantial improvement in network cell capacity.
 

Futures and challenges facing beamforming technology

In common with many other areas of 5G networks, developers of antenna systems must meet the twin demands of ever-shrinking components and reduced power consumption.

The pressure to increase spectral efficiency and throughput is leading to the specification of ever-larger antenna arrays, with 64 x 64 MIMO, and larger already on the horizon. The effectiveness of beamforming is heavily dependent upon the precision of the antenna arrays, with the strength of unwanted side lobes increasing as the spacing between elements approaches the signal wavelength. At 60 GHz this wavelength is 5mm, giving some idea of the manufacturing tolerances required.

Shrinking wavelengths also mean shrinking components, such as RF transceivers, which must integrate RF power amplifiers with functionality such as ADCs. At the same time, designers must find ways of improving power efficiency of all 5G network components. RF power amplifiers for mmWave have traditionally been the preserve of III-V semiconductor materials such as GaAs. However, these devices are not sufficiently power efficient, and do not integrate well with other functionality. Advances in 40 nM CMOS are therefore welcome, enabling the size and power consumption of these key components to shrink further.

Also, as more beams are generated by individual gNBs, the signal processing requirements become more complex. Research and development into areas such as beam synchronisation is ongoing, with neural network techniques being deployed - requiring advanced processing hardware, further stretching power budgets, and adding space constraints.
 

Conclusion

The 5G promise relies on a successful roll-out of mmWave technology. Both MIMO and beamforming are critical components, enabling the capacities and throughputs required by emerging applications and exponentially growing numbers of IoT devices.

In a few short years, MIMO and beamforming have moved out of the research environment into commercial deployment - first in LTE networks and now in early 5G deployments. 3GPP specifications place heavy demands on the continued evolution of these twin functions, and ongoing developments in supporting technologies such as Advanced Antenna Systems (AAS), 40 NM CMOS and software processing pave the way for ever larger MIMO arrays.