Looking at 5G from a microwave perspective

5G represents both an evolution and a revolution of mobile technologies, to reach the various high level goals that have been published to date by the various members of the wireless ecosystem. Here, Thomas Cameron, CTO, Communications Infrastructure, Analog Devices explains.

5G is widely seen as the generation of wireless that will enable cellular to expand into a completely new set of use case and vertical markets. While 5G is generally seen as the technology to deliver ultra-broadband services including HD and ultra-HD video streaming, 5G technology will also enable cellular to enter the world of machines, contributing to autonomous vehicles, connecting millions of industrial sensors and a multitude of wearable consumer devices.

The evolutionary path to 5G consists of incremental enhancements of 4G in the conventional cellular bands and extending up in frequency to emerging bands in the 3- to 6-GHz range. Massive MIMO (multiple-input and multiple-output) has industry momentum and will evolve from the first systems based on LTE to adopt new waveforms designed to improve throughput, latency and cell efficiency.

Spectrum is seen as the lifeblood of the cellular industry and the spectrum in the legacy cellular bands (sub 6GHz) just cannot support the exponentially growing demand in the coming years. As such the bands above 6GHz are currently under study, to test the viability of deploying wireless access in frequency allocations above 6GHz. While the collective global spectrum available below 6GHz is on the order of hundreds of MHz, the amount of potential spectrum above 20GHz is in the tens of GHz. The taming of this spectrum is considered essential to achieve the 5G vision of a truly connected world.

A higher level of operation

As a result, a segment of 5G is likely to operate on much higher frequencies (possibly up to millimetre waves) and will likely adopt new air interface technologies that are not backward compatible to LTE. The frequency bands discussed among key industry players include higher frequency bands such as 10, 28, 32, 43, 46 to 50, 56 to 76 and 81 to 86GHz. However, these bands are currently in the proposal stages and much work remains to be completed in channel modelling prior to the radio systems definitions and standards deliberations. The ITU recently published a plan for 5G standardisation with a target to publish the first generation of IMT-2020 specifications around year 2020.

Obstacles to deployment

Given that 5G is still in its infancy, work also remains relative to channel modelling, radio architecture definition and chipset development before the first commercial systems will be deployed. However, there are certain trends and requirements already agreed upon and problems to be solved that will lead to final 5G systems.

Let’s consider 5G access systems at microwave and mmwave frequencies. One of the major hurdles in implementing radio access at microwave frequency is overcoming the unfavorable propagation characteristics. Radio propagation at these frequencies is highly affected by atmospheric attenuation, rain, blockage (buildings, people, foliage etc.) and reflections. Microwave point-to-point links have been deployed for many years, but these are generally line-of-sight systems. The fact that they are stationary makes the link manageable, and systems have been developed in recent years that support very high throughput using high order modulation schemes. This technology continues to evolve, and we will leverage the microwave link technologies into 5G access.

Early in the cycle, it has been acknowledged that adaptive beamforming will be required to overcome the propagation challenges for access systems. Unlike point-to-point systems, beamforming will need to adapt to users and the environment to deliver the payload to the user. It is generally agreed in the industry that hybrid MIMO systems will be used in the microwave and low mmwave bands while in V and E bands, where bandwidth is plentiful, the systems will likely only employ beamforming to reach the required throughput goals.

The high level block diagram below depicts a hybrid beamforming transmitter. The receiver can be envisioned as the reverse. The MIMO coding is performed in the digital section along with the typical digital radio processing. There may be a multitude of MIMO paths processed in the digital section from the various data streams feeding the antenna system. For each data stream, the D/A converter converts the signal into analogue at either a baseband or IF frequency depending on the selected architecture. The signal is upconverted and split into the constituent RF paths to feed individual antennae. In each RF path the signal is processed to set the gain and phase to form the beam out of the antenna.

While the block diagram is simplistic, the system challenges and trade-offs are complex. In this short treatment of the topic, only a few issues will be discussed, but let’s focus on the architecture and radio challenges. It is critical to design this system with power, size and cost in mind from the start to bring these systems to reality.

While such radios can and are being built today for prototype 5G systems using discrete (mainly GaAs) devices, we need to bring the same high levels of integration to bear in the microwave space as what has been implemented in cellular radios. High integration and high performance create a tough problem for the industry to solve.

However, integration alone is not the solution to this problem facing the industry. It requires smart integration. When we think of integration, we need to first consider architecture and partition to leverage the benefits of integration. In this case mechanical and thermal design also need to be considered as the circuit layout and substrate are interrelated.

First of all, an architecture conducive to integration needs to be defined. If we consider the examples of highly integrated transceiver ICs for cellular basestations, many use a zero IF (ZIF) architecture to either eliminate or minimise the filtering in the signal path. Particularly at microwave frequency one must minimise the loss in the RF filters, as RF power is expensive to generate. While ZIF will reduce the filter issue, of course the trade-off is LO suppression, but we shift the problem from physical structures to signal processing and algorithms. Here we can leverage Moore’s Law, whereby passive microwave structures do not follow the same scaling dynamics. It is necessary take advantage of the ability to optimise analogue and digital simultaneously to reach our goals. There are many algorithms and circuit techniques that have been employed at cellular frequency that may benefit the microwave space.

Next, consider the semiconductor technology requirements. As mentioned above, state of the art microwave systems are generally implemented with GaAs components. GaAs has been the mainstay of the microwave industry for many years, but SiGe processes are overcoming the barriers of high frequency operation to rival GaAs in many of the signal path functions. High performance microwave SiGe BiCMOS processes enable a high level of integration required for these beamforming systems encompassing much of the signal chain as well as auxiliary control functions.

GaAs PAs may be required, depending on the output power required at each antenna. However, even GaAs PAs are inefficient at microwave frequency as they are generally biased in the linear region. Linearisation of microwave PAs is an area ripe for exploration in the 5G era, more than ever before.

CMOS

What about CMOS? Is it also a contender? It is well documented that CMOS is well suited for high volume scaling and is being proven out in WiGig systems at 60GHz. Given the early stage of development and uncertainty of the use cases, it is difficult to say at this point if or when CMOS will be a technology choice for 5G radios. Much work needs to be done first in the area of channel modelling and use cases to come to a conclusion on the radio specifications and where microwave CMOS may fit in future systems.

The final consideration of 5G systems is the interdependency of the mechanical design and RFIC partitioning. Given the challenges to minimise losses, the IC needs to be designed with the antenna and substrate in mind to optimise the partition. Below 50GHz, the antenna will be part of the substrate, and it is expected that the routing and some passive structures may be embedded in the substrate. There is a body of research ongoing in the area of substrate integrated waveguides (SIW) that looks promising for such integrated structures. In such a structure it will be possible to mount much of the RF circuitry on one side of the multi-layer laminate and route to the antennae on the front face. The RFICs may be mounted in die form on this laminate or in surface mount packages. There are good examples in the industry literature of such structures for other applications.

Above 50GHz, the antenna elements and spacing become small enough that is it possible to integrate the antenna structure in or on the package. Again, this is an area of ongoing research that may push 5G systems forward.

In either case, the RFIC and mechanical structure must be co-designed to ensure symmetry in routing and to minimise losses. None of this work will be possible without powerful 3D modeling tools for the extensive simulations required for these designs.

While this is a brief perspective on the challenges 5G brings to the microwave industry, there will be boundless opportunities to bring forth RF innovations in the coming years. As mentioned, a rigorous systems engineering approach will yield the optimum solution by leveraging the best technologies throughout the signal chain, developing processes and materials, creating design techniques and modeling and defining high frequency test and manufacturing. All disciplines have a role to play in reaching the 5G goals.

It’s an exciting time to be an RF engineer in the wireless industry. 5G is only just starting and there is much work ahead in order to realise commercial 5G radio networks by 2020.