元器件 > 高级半导体博客



5G and the Election?

by Eric Higham | 10月 31, 2016

The United States is nearing the end of a very contentious presidential election cycle. Political commentary has long embraced military terminology, so we hear of battleground states, balance of power, boots on the ground, brinksmanship and damage control. Given the fervor of the claims and criticisms from both sides, I am reminded of the term “Pyrrhic victory”, where a victory inflicts such a massive toll that overall defeat is virtually assured.

Politics aside, it’s always important to keep the objective in mind. It’s easier than it seems to have an altruistic goal, but forget that a scorched earth policy of achieving that goal probably doesn’t get you the best result. If you haven’t been faced with that situation, consider yourself lucky. It’s essential that you don’t make the process more important than the result.

While that advice pertains to political, military, emotional and intellectual decisions, it also pertains to technology and 5g (one of my favorite topics). As the race to 5G deployment intensifies and companies jockey for “first mover advantage” to establish themselves as leaders, we are starting to see battle lines being drawn ( Verizon rejects AT&T-led effort to speed up release of parts of 5G standard). Earlier I said don’t let the process become more important than the result and in the case of 5G technology; the process is “the process”.  We are seeing compound semiconductor processes like GaAs, GaN and SiGe vie with RF CMOS for their rightful place in the 5G RF application space.

This promises to be a fascinating battle as antenna technology will be one of the key enablers for the vision of 5G. The 5G networks will make use of “Massive MIMO” and while there is no universal definition for what this means just yet, these antennas will likely contain at least 64 elements, be able to generate multiple simultaneous beams, make use of multipath propagation techniques with beams that are steerable in both azimuth and elevation. All of these techniques are necessary to achieve the data rates and capabilities promised by the 5G vision. This is where the fun…and uncertainty starts.

Beam steering/beam forming and the MIMO (Multiple Input Multiple Output) capabilities rely on antenna element spacing that makes an array large (and problematic for user equipment) below 6 GHz, but this will be the first approved 5G frequency band. There is a substantial amount of effort at millimeter wave frequencies where the element spacing becomes much convenient for equipment, but the propagation characteristics and the component issues become more challenging. The millimeter wave frequencies for 5G will not be officially ratified until WRC19 in 2019 and this portion of the 5G network is expected to lag initial rollouts by two years. The lack of officially ratified frequencies has not stopped the industry and network results at 15 GHz and 28 GHz have recently been demonstrated.

 

Massive MIMO architecture
Source: Qorvo

In addition to frequency, antenna architecture is a fertile battleground. The graphic above from a Qorvo presentation at the 2016 Compound Semiconductor International Conference shows the advantages and disadvantages of the different approaches under consideration. The “All Digital BF” version does all the beamforming in the digital domain and amplifiers drive small arrays of elements. This provides the most flexibility for the array and the RF output power is low, but this architecture demands a lot of performance from the silicon baseband devices and increases the DC power consumption. At the other extreme is analog beamforming, or “Switched Beam”. This approach uses fewer baseband devices, but it does beamforming in the RF domain, it relies on higher power amplifiers and this architecture increases the complexity of the antenna. In addition, you don’t get all the advantages of multipath propagation. The approach that seems to be gaining a lot of acceptance is the “Tiled Hybrid BF” approach in the center of the graphic. This approach represents a compromise with the other two approaches.

This uncertainty multiplies as we introduce the process technology variable. GaAs, GaN, SiGe and RF CMOS all have advantages for the different architectures under consideration. With today’s silicon baseband technology, there is a sweet spot with power dissipation versus the number of elements. With fewer elements, each one needs a higher transmit power and GaN or GaAs is likely a better choice. The RF transmit power levels of an all-digital beamforming architecture reduces the transmit power to levels that SiGe or RF CMOS can handle. The other variable is the advance in performance of CMOS technology used for the baseband/processing chips. As the silicon performance changes, it will change the sweet spot for the architecture and process technologies.

So, there are many dimensions to the development of 5G networks and the opportunity is enormous. As the large RF companies continue to consolidate, they are becoming more technology agnostic and less likely to have a vested interest in one specific technology. As I said in the opening, 5G success will go to companies that keep the goal in mind and don’t make the process more important than the result.

 

Have a safe and happy Halloween and remember to vote!

  • Eric
Previous Post: Is the GaAs Device Market Running out of Gas? | Next Post: Is DOCSIS 3.1 Bad for the RF Industry?
Leave a comment