Direct line of sight (LOS) is a luxury in wireless communications, confined largely to some dense urban environments and some mostly legacy fixed wireless access (FWA) services in relatively flat treeless settings.
Yet LOS was initially thought to be critical for long distance transmission in millimeter wave frequencies above 20 GHz, in which range of up to 10 kilometers (6 miles) has now been demonstrated. In practice, the range of mmWave connections are limited to 500 meters or less in urban settings and that has been a factor limiting the spectrum’s utility. It has also been a reason why spectrum at those high frequencies has attracted far lower prices at auction than sub-GHz and midband spectrum.
Indeed, the range limitations are why some operators have spurned mmWave altogether so far.
It became clear that, for mmWave to deliver on its earlier promise of enabling ‘true 5G’, leveraging the high capacity available to deliver very high bit-rates, progress had to be made on making it enable better coverage even when the operator could only achieve near-line of sight (NLOS).
This is now being achieved, in part because of belief and support from leading vendors, especially Qualcomm, which has been a strong advocate of mmWave and been developing NLOS transmission technology for several years now. At a recent webinar Qualcomm set out its stall as a leader in silicon implementation of beamsteering, which is essential for extending mmWave range in environments where trees, buildings and small hills impede the signal path.
Beamforming comes in several guises but is, in any case, essential for overcoming signal attenuation through the air at those high frequencies by focusing the energy rather than splaying it out in every direction as happens for RF emissions normally. Fortunately, at higher frequencies the shorter wavelength enables use of smaller antennas, which means more can be accommodated in a device. That in turn increases scope for beamforming by enabling signals to be focused into even sharper beams.
This is possible because with RF radiation, as opposed to visible light or sound, the waves can be focused by exploiting constructive and destructive interference between neighboring emissions. By offsetting the phase among multiple signals, the interference can be tuned to be constructive along the desired beam path and destructive off it.
Beams are usually focused by having multiple antennas close together broadcasting the same signal at different specified times so that they are slightly out of phase. Antennas can then be arranged in clusters at a base station, each cluster focusing beams tightly in a given direction. But when beams are blocked by a physical object on the way, they are reflected or bent, and arrive offline, slightly out of the intended phase.
Fortunately, the existence of multiple paths can be turned to advantage by providing some redundancy – if a beam cannot get through one way it may well by another. However, this poses a computational challenge because complex signal processing is required to recover each beam accurately at the receiving end.
Beamforming enables spatial multiplexing whereby the same time and frequency slot can be used to send different information in separate beams, increasing capacity. When different data streams are sent this way to just one receiver, the result is Single User MIMO (SU-MIMO), while when streams are transmitted to more than one device it is called Multiuser MIMO (MU-MIMO). Massive MIMO involves high antenna density (at least 16T16R), such that a larger number of beams can be sent and received by individual devices, which effectively confines it to base stations rather than user equipment. Up to 128 antenna elements have been demonstrated – or even 256 in lab trials – limited by the minimum antenna size at a given frequency, often around 4 centimeters for mmWave.
With NLOS the ability for user devices such as smartphones to steer beams becomes desirable or essential to ensure optimal transmission and resiliency. This minimizes the varying impact of signal loss during transmission as a user moves around, which changes the RF landscape between device and base station. Without beamsteering there are various losses, for example from variations in delay with frequency across the wider mmWave bandwidths, which leads to shifts in the direction the beam is pointing.
Fortunately, the accuracy of beamsteering in mmWave from smartphones does not have to be exceptionally high, with an angular range of about 60 degrees being adequate to enable sufficient spatial coverage and focusing of peak Effective Isotropically Radiated Power (EIRP).
Beamsteering then does not require any physical manipulation, being accomplished for RF signals either by switching between antenna elements with different configurations, or else more efficiently by changing the phases of signals appropriately to alter the interference patterns.
For this to work there has to be some way for devices to determine the amount of steering required and adjust this in real time, which imposes computational complexity. This can be done in a four-step process that begins with beamsweeping, where beams are periodically transmitted in all predefined directions in a burst, to be sure devices can perceive them at some point. Then the quality of each beam’s received signal is measured.
Then, the device then selects the most optimum beam giving the best signal at that time and sets up a directional communication. Finally, the device reports which beam it has selected.
Qualcomm has made efforts to reduce the complexity involved so that beamsteering can be implemented in its chips for mobile devices. The challenge is that under earlier proposed systems, both signal gain and phase had to be controlled separately in real time for each antenna, which becomes computationally prohibitive for a large array.
Qualcomm has shown that it is possible to deploy beamsteering effectively in practice just by controlling the phase, greatly reducing the computational toll. The caveat is that large signal transmission losses can then happen, but these only occur in a small minority of channels, so can be tolerated statistically. Qualcomm has as a result declared mmWave NLOS beam steering fit for deployment.
We should note this is not the last word in mmWave NLOS beamsteering, as research is still ongoing, with scope for further improvements in cost, capacity and coverage. For example, the UK’s University of Birmingham in June 2022 presented experimental results of a radically different antenna type for mmWave frequencies.
Rather than relying on phase-shifting, this found a way of physically shifting the directions in a device the size of a smartphone made from a metal sheet punctured with many evenly spaced holes only a few micrometers in diameter. In each hole is a cavity whose height is controlled by an actuator to make movements on a micrometer scale that deflects the beam of an emitted radio wave from that hole, redirecting the energy physically.
Whether this would be cost-effective or reliable in sustained operation remains to be seen, but it shows there is plenty of mileage left in beamsteering research.