It has been a bad week for 5G as rumors leak out that Verizon is struggling to sell its “5G Home” fixed broadband replacement service while an event in Hawaii billed as the first real world test for mobile 5G flopped dismally. These could be said to be teething troubles or little local difficulties, but still underline a wider sense that 5G is being misrepresented and rolled out too quickly, taking precedence over more pressing issues of coverage, price and consistency.
We have been critical of 5G in the past, not about the underlying technologies, but the false premises and promises. It is more of a roadmap than a single “G”, comprising a range of technologies under the banners of higher bandwidth and lower latency. There is an implicit assumption of essential infrastructure elements, especially faster backhaul and smaller cells, together providing the increased capacity which does not just appear magically out of the spectrum. It has to a large extent been hijacked both by operators desperate to reverse falling revenues or customer numbers and the infrastructure vendors such as Ericsson and Nokia seeking ammunition against their competitors. As an aside, those two have been boosted by Huawei’s recent travails around the arrest in Canada of its founder’s daughter, which left Ericsson stocks as the sole gainers among major tech players in Europe. That is probably just a temporary gain.
Verizon’s 5G came on our radar screen three months ago mid-September with the announcement of 5G Home based on its so called Ultrawideband 5G network, with access initially confined to the four US metropolises of Houston, Indianapolis, Los Angeles and Sacramento. As we pointed out, the sheen of being first to something approaching 5G would wear off fast and that looks like it has been the case. The principle problem as we also underlined was that the 300 Mbps service priced at $50 a month for Verizon and $70 for non-Verizon wireless customers would be likely on average to cost more than the guaranteed service at the same speed Verizon already sold on fiber to non-wireless subscribers.
We questioned whether this would entice half a million customers to throw out a perfectly good Comcast DOCSIS 3.1 line that actually delivers 1 Gbps all the time.
Apart from being confined to fixed broadband replacement, Verizon’s boast of being first to launch a commercial 5G service was a stretch because it was based on an earlier pre-standards version called 5G TF. There was a vague pledge in the near future to upgrade to 5G NR, which is generally accepted as the more definitive version. This has left Verizon talking about its first 5G service available on the move and therefore worthy of the name mobile arriving in the spring of 2019. In that case, AT&T will almost certainly be first out with a mobile offering based on 5G NR, although it has suffered from teething troubles of its own and delays after promising to launch 5G in 19 cities sometime towards the end of 2018 based on infrastructure from Ericsson, Nokia and Samsung. That does not leave long.
Meanwhile, both Verizon and AT&T, along with these vendors and others, received a setback at the Qualcomm Snapdragon Technology Summit early this month on the Hawaiian island of Maui in an event billed as the coming out party for full speed 5G, before the cellular standard is launched formerly in the US. It was supposed to be the first scaled up real world trial featuring equipment from the leading vendors, but it underperformed so badly none of the participants attempted to disguise the fact. Instead they tried to save face by disingenuously claiming it was just to demonstrate use cases rather than speeds.
But it was supposed to be all about the speed, as well as low latency. The main concern was that while Qualcomm’s newly announced Snapdragon 855 processor featuring in the equipment is certainly capable of multi-gigabit speeds, and its predecessors over LTE networks have already crossed the 400 Mbps barrier, in this case the single 39 GHz millimeter wave 5G network only ran at 140 Mbps. This was according to Ericsson, which set up the trial network for both Verizon and AT&T, but then said that the 100 MHz spectrum provided was not enough, especially with so little time to test and optimize it. Latency was also disappointing, although again that would be improved with more spectrum.
Given that the trial was misconceived from the start, we should not make too much of it, but it does nothing to advance the cause of 5G. It rather corroborates the arguments advanced in a little book published two years ago by Professor William Webb called the 5G Myth: And why consistent connectivity is a better future. He argued that the urgency to deploy 5G was driven entirely by commercial considerations when consumers and enterprises would be much better served by a focus on consistency and coverage.
It has also become clearer since that book was published that 5G as currently enshrined in standards does not cater enough for the wide range of use cases for which it is envisaged, which is odd given that it is being promoted partly for its greater versatility. This is particularly true for latency, where the consensus seems to be that 5G needs to be faster than 4G but within manageable bounds, with round trip delays somewhere in the range 1 to 100ms. But consideration of proposed cases indicates a much wider range of latencies covering up to perhaps 10 orders of magnitude according to one recent study.
On the one hand there will applications under the IoT banner where neither latency nor bit rate are that critical, such as sensors for predictive maintenance, feeding back data at a trickle that might trigger a maintenance visit a week earlier than planned when some tipping point is reached. At the other extreme, some network timing devices might require responses to be delivered in nanoseconds.
Then there are many use cases in between, occupying varying points on the latency demand spectrum, such as temperature sensors, thermostats and security tags.
In the case of 5G, most of the focus has been on the role of edge networking to keep latency within reasonable bounds, especially in the context of QoS for video. Here, a target of 1ms has often been cited, but that is overkill for many other use cases, while not being fast enough for some, where 5G would not be good enough in any case, calling for direct fiber connections or even on-chip connectivity. The main point is that not enough attention has been paid to the performance spectrum and spread of demand across it, which is perhaps asking too much.
Still, the good news is that there is plenty of time for 5G to evolve and adapt to demands, because it will arrive more gradually than the likes of Verizon would like. It will be years before we can talk about an all 5G network and it will arrive on a patchwork basis and critical applications will require fallback to 4G or even WiFi precisely because of poor coverage. There are signs of this already, not so much in Verizon’s 5G Home which does not meet a pressing demand, but perhaps in Vodafone’s prospective commercial 5G services slated for launch in mid-2019. The operator provoked some surprise when it announced that this would be available in some of the UK’s most rural areas in Cornwall and Cumbria as part of a plan to provide much better coverage in remote regions by leapfrogging straight to the latest cellular technology.
Such use cases, including fixed-wireless broadband, will play a part but we do not anticipate that mounting much challenge to fiber or even copper/coax at large scale because it will not make economic sense. After all, 5G small cells will need plenty of fiber backhaul anyway, so they might as well be used to provide direct high-speed connectivity in the fixed broadband scenario.
In the end, the technologies of 5G will be driven more by urban use cases and demand for greater capacity to serve video in particular to smartphones, exploiting the increased cell density along with greater antenna efficiency and higher modulations.