Technology and methods for providing beam-forming antennas for 5G mobile communications systems
5G communications standards promise to enable a thousand-fold increase in wireless data capacity over the next ten years.
But with the amount of data that can be coded onto a single channel approaching theoretical limits, it will take a combination of frequency, time and spatial multiplexing to create the multiple channels necessary to get data from sender to receiver at such high aggregate rates for 5G mobile communications systems as well as other wireless systems.
What are our options? We can borrow from the MIMO (multiple input, multiple output) techniques used in WLAN. We can take advantage of new frequency bands, including some at millimeter wavelengths, that regulators will make available.
And we can develop techniques to steer mm-wave signals to create a more direct link from sender to receiver, to counter their strong attenuation in free space.
Tackling path loss with more antennas
Why is there such an attenuation problem at millimeter wavelengths? Consider a communication link between a base-station and a smartphone, operating at mm-wave frequencies, in which the smartphone has an isotropic antenna (that is, one which radiates and receives equally in all directions). The path loss between the transmit and receive antennas is given as in :
&nbps; where PRX and PTX are the received and transmitted powers, and GRX and GTX are the receive and transmit antenna gains, respectively. The path loss is partly due to the attenuation caused by the signal’s energy spreading into an increasing volume as the distance R between transmitter and receiver grows, and is defined by 1/4πR2. The other key factor in path loss is the amount of energy that the receive antenna can capture, which is controlled by its effective aperture (defined by the λ2/4π term) and shrinks with the square of the wavelength.
This means that, for example, changing the signal frequency from 3GHz to 30GHz (and therefore shortening the signal’s wavelength) increases the path loss by 20dB. To compensate, we can increase the number of receive antennas, but it takes 100 antenna elements receiving a 30GHz signal to achieve the same total array aperture and therefore received power as the original antenna at 3GHz.
Beam forming in antenna arrays for 5G communications
How do you build such an antenna array? The simplest form consists of N elements spaced closely together at regular intervals, a distance d apart (Figure 1).
Figure 1: A one-dimensional antenna array, in which all elements have identical phase, points its beam towards θ = 0 degrees
If all the elements in such an array are isotropic, have the same gain, and are driven with a signal at the same phase and power, the resultant beam will point straight out of the plane on which they are mounted (i.e. in the z direction). The resultant field is shown in Figure 2 as a function of θ, the angle between the z-axis and the observation direction, when the distance between the array elements d is λ/2 (half the wavelength).
Figure 2: Side-lobe patterns for a one-dimensional antenna array in which all elements have identical phase, creating a beam at θ = 0 degrees. The array consists of 64 elements.
If we apply a phase difference between adjacent array elements, the beam can be directed towards another angle, for example 20 degrees, as shown in Figure 3.
Figure 3: The same array, but with all elements shifted in phase to create a beam at θ = 20 degrees
In both cases, unwanted side-lobe signals form alongside the main beam. If the array elements are spaced more widely, the strength of the side lobes increases until, when the separation distance d matches the signal wavelength λ, unwanted beams with the same power level as the main beam appear at +90 and -90 degrees. In Figure 4, the separation distance d is twice the signal wavelength, and unwanted lobes have been created at ± 30 degrees and ± 90 degrees. These lobes are usually unwanted, since they increase the likelihood of one antenna array interfering with another.
Figure 4: Grating lobes appear in the array radiation pattern when the inter-antenna distances exceed the wavelength, in this case d = 2λ
In practice it is not possible to make an isotropic antenna, since all physical antennas have a certain antenna pattern, and so the ability to steer them electrically is limited.
Analogue and digital beam forming
How do we create the phase shifts necessary to steer the beam? Analog beam forming can be done in the RF domain by using phase shifters in front of each antenna, as in Figure 5.
Figure 5: In analogue beam forming, the beam is steered using phase shifters. Only one data stream and one beam can be generated
A single data stream is handled by a set of data converters and a transceiver. After the transceiver, the transmit data stream is split as many ways as there are array elements. The signal in each branch passes through a phase shifter, is amplified and then fed into the array element.
Analogue beam forming in the RF path is simple and uses a minimal amount of hardware, making it the most cost-effective way to build a beam-forming array. The drawback is that the system can only handle one data stream and generate one signal beam.
Digital beam forming, in which each antenna has its own transceiver and data converters, can handle multiple data streams and generate multiple beams simultaneously from one array, as shown in Figure 6.
Figure 6: In digital beam forming, the beam is steered by baseband processing. Multiple data streams and beams can be generated simultaneously
The phase differences needed to generate a signal beam are created in the baseband, which can also create several sets and superimpose them on the array elements. This enables one antenna to generate multiple beams, each with its own signal and serving multiple users, with one array and one set of spectrum resources. This approach needs more hardware and puts a greater burden on the signal processing in the digital domain than the analog approach.
Beam forming, massive MIMO and channel state information for 5G
Digital beam forming can be used for 5G mobile communications and can point a signal from a sender to a receiver when they are in line of sight. When they aren’t, users are only reached by beams that have been scattered by buildings, trees and other features of the environment.
Scattered beams suffer 20 to 30dB more path losses than line-of-sight beams, so it makes sense to use many of them to ensure that the sum of their scattered signals at the receiver provides enough energy to correctly interpret the communication they are carrying.
Such low-power, scattered beams only interfere constructively at the user’s location, with other users nearby only experiencing their signals as a slight increase in background noise. This means that an antenna array can serve several users, each with a multitude of beams that are being scattered, so long as the number of elements in the array exceeds the number of users. This is known as massive MIMO.
It is possible to take massive MIMO a step further for 5G mobile communications by replacing an antenna array of N elements with N individual antennas distributed widely through the environment on separate buildings, lamp posts, etc. Applying time delays to the distributed antennas ensures that signals for each user only interfere positively at their location.
As seen above, an antenna array with element spacings less than the signal wave length only radiates very little energy in unwanted directions. This isn’t so for a set of widely distributed antennas, whose signals can be configured to interfere constructively at the user’s location, but which do not interfere destructively in other directions and so may cause interference and power loss.
The information necessary to work out the phase shifts needed to form beams can be derived from the channel state information. One way to obtain this is by having the user equipment transmit a pilot tone and then setting the bases-station to measure the phase shifts between the various paths that the signal follows between the user equipment and the base-station’s antenna elements.
For this to work, the path from the sender to the receiver must be the same as the path from the receiver to the sender, which means that they must be at the same frequency and hence that the overall system must use time-division duplexing. And phase shifting the signals between antennas can only compensate for the different path lengths at one frequency. For large signal bandwidths, the number of pilot frequencies should increase to map the channel behaviour properly.
We can conclude that digital beam forming offers the most versatile solution for future 5G communication systems, but is also the most expensive and complex implementation. It can therefore be expected that the first 5G mobile systems will use some kind of combination of analogue and digital beam forming, such that proper trade-offs can be made between system performance and cost.
 C.A. Balanis, Antenna Theory, Analysis and Design, 3rd ed. 2005, John Wiley & Sons, ISBN: 0-471-66782-X