In our corner of the tech world there is an endless tempo of 5G ‘excitement’. This is largely manufactured because, as we wrote about last March, 5G brings no immediate, tangible gains for consumers. That has not stopped the marketing engines at many companies from pushing the topic for all they are worth.
To be fair, 5G is important. Every new ‘G’ in mobile brings meaningful improvements, but it is going to take a while to roll them out this generation. We raise this topic because in recent weeks we have been feeding a steady stream on inbounds from all sorts of clients and investors asking about 5G. (Which is ironic because our family members were asking about it in March, putting them way ahead of the pros…).
So here we want to address the big issues in 5G. For some background, please see our posts from Mobile World Congress. A quick recap:
There are really two phases of 5G. The current phase, which we call sub 6 is really just an evolution of 4G; modest improvements in data rates as well as some important, but hard to observe, changes in the software the operators use to run their networks. The next phase, which we call mmWave, will bring much more tangible changes, notably including data rates at or above 1 Gbps.
In a lot of the discussion about the ‘Race to 5G’, this distinction gets lost. Many carriers are rolling out sub-6 deployments, with China and the US leading the pack, but no one is deploying a mmWave network for widespread commercial use yet. We argue that mmWave is, in fact, still pretty far away.
By our count, there are 5 key challenges facing mmWave, three technical and two non-technical non-technical:
- Heat and device cooling
- Radio Frequency chips (RF)
- Networking software
- Network densification (aka capex $)
- Small cell siting (aka zoning regulations)
We discussed the subject of Device Heat in a post last week. We want to add here that everyone new generation of wireless standards has seen heat issues with new phones, but those were solved quickly. The problems with 5G mmWave are larger and will not go away as quickly. Handset makers are just waking up to the existence of this problem.
Every advance in the standards poses new challenges to makers of RF components. With mmWave, RF suppliers are going to need a whole new skill set, as the fundamentals of the problem are very different. By now, those following RF are familiar with all the parts of an RF front-end including power amplifiers and filters, (This primer on the subject from 2015 remains one of our most read posts.) but it’s possible that mmWave RF will need entirely new materials. Not a small change.
Qualcomm has announced a complete, working mmWave RF front-end, that is sampling now. But to our knowledge, at this time no one else is even close to having a working product. And if you read that Qualcomm press release closely you will see that it is not a mobile solution, it is targeted at Fixed Wireless Access, so a box that sits in your window. A totally different form factor and set of requirements. To be fair, this is a pretty impressive product, full credit to the Qualcomm team for shipping this, but shrinking all of that to a mobile form factor will not be an easy task. Moreover, it is unclear how big the market for mmWave can be until there is more than one supplier of RF products.
Networking Software. An important, but unsung component of every ‘G’ is the way in which the base stations manage all the phones with which they connect. This was a thorny issue in 3G, and the competitive landscape was altered dramatically by the way in which it was solved (raise your hand if you remember make before break and hard vs soft handover). The signalling overhead in wireless communications can have a large impact on user experience. Poorly implemented, it will lead to reduced bandwidth, dropped calls and a host of other problems.
An important aspect of every ‘G’ in wireless is that it brings better ‘spectral efficiency’, meaning the operators get more from their spectrum assets. Crucially, that is not true in 5G. As we roughly calculated in this tweet, 5G is at best 15% more efficient than 4G, and quite possibly less efficient in deployment. If the routing software imposes some further penalty, then the industry has a real problem. All of this is solvable, but a lot of that work needs to be done.
mmWave will require a LOT more ‘base stations’ because these frequency bands do not travel very far. We’ve heard estimates of mmWave requiring 10x as many ‘small cells’ as 4G has macro cells (i.e. base stations, note here we are just going to call mmWave base stations small cells and everything else a base station). Coordinating all that activity has long been viewed as beyond the capability of current base stations technologies. This thorny issue touches on problems of contention and cell-to-cell interference. A good clue that someone has worked out a solution for this will be the use of “Artificial Intelligence” in their marketing language.
This topic of network densification brings us to the two non-technical challenges. First, someone is going to have to deploy all those small cells. Even if the boxes themselves only cost a few hundred dollars, the number required to support reasonable mmWave coverage quickly adds up to billions of dollars. Having just shelled out for new spectrum purchases the operators are going to be cautious about embarking on massive capex plans. One of the great things about 4G, from the operators’ perspective was the much simpler upgrade path for their 3G base stations. Again, this problem is solvable, but to enable reasonable financial burdens, many (most) telecom operators are adopting a go-slow deployment model, at least for now. Alternatively, this may open up new economic models for small cell deployments (and if we really want to go out on a limb, this is a really intriguing use case for blockchains).
Separate from the financial problem will be where exactly will all those small cells be located. To cover a city, mmWave small cells will need to be affixed to light posts or buildings on ‘every’ block. This will be a messy process requiring operators to negotiate with local governments and property owners at a block-by-block level. It took the better part of 20 years to get the coverage we enjoy today, and now we are talking about a huge increase in the number of sites required. To make matters worse, the operators cannot rely on traditional middlemen such as US tower operations REITs like Crown Castle, as they themselves do not have these sites in their portfolio. In our view, this siting issue is the single biggest obstacle to mmWave deployment.
From all of this, it should be clear that mmWave is still many years away. After watching feedback from some recent conferences on the subject, we have come away with the impression that by the time many of these issues are worked out, the industry will already be ready for 6G. We would not be surprised if mmWave deployments end up getting bundled into the next generation, and we just end up calling mmWave 6G. As we noted in our piece from Mobile World Congress, the operators only make money on the even-numbered G’s anyway.
UPDATE: After we wrote this post, but before we published Qualcomm hosted a 5G event talking about their view on challenges to 5G mmWave deployments. We did not attend the event, but found this report from DigiTimes provides a good overview.