5G was the dominant theme at this year’s Mobile World Congress, and the news has continued with the announcement last week of an accelerated standards timeline for a non-standalone “5G” specification. This development, and operators’ commitment to launch a standards-based mobile 5G service as soon as late 2018, tells us three things about the 5G rollout: It will occur in stages; it will be messy; and there will be multiple “versions” of 5G.
Let’s remind ourselves of the original vision of 5G. I went back to some vendor slideware from a couple of years ago, and some of the 3GPP materials, which promised the following key capabilities for 5G:
- Speeds of up to 10 Gbps, with the ‘objective’ of 1 Gbps
- Latency of 5 milliseconds (ms), with <1 ms objective for some applications
- Data capacity of 1,000x per square km, compared to LTE(!)
- Ability to handle up to 1 million devices per square km (IoT)
- Use of wideband spectrum, 200 MHz or greater, and use of millimeter (mm) wave bands for short wireless links
- Massive MIMO (64+ ports)
- NFV and SDN key to the development of a unified core solution
The vision of where we want to be with 5G hasn’t changed. What has changed is the industry’s willingness to have a few flavors of 5G, and move to it in stages. The intensified competitiveness, growth in data traffic (not matched by similar revenue increases), and a lot of hard work by many talented people all mean that operators will deploy elements of 5G that represent significant enhancements, but won’t fulfill all aspects of the 5G vision, at least initially. Not that there’s anything wrong with that.
What does this look like, as of mid-2017?
- Test 5G. These are the mainly fixed wireless access trials being conducted by Verizon and to a lesser extent AT&T, this year. There are no guarantees this will be interoperable with the finished 5G standard.
- 5G Non-standalone option: This is what was approved last week. Will require an anchored LTE, using 5G new radios to deliver faster network speeds and lower latency. Services possible by late 2018
- 5G Standalone option: Would not require an LTE anchor. 5G-specific core. Likely greater use of NFV & SDN, and network slicing. Standard completion 3Q/4Q 2018, services possible 2019
- Release 16. This is the more ‘fully formed’ version of 5G, closest to the original 3GPP vision: support for massive # of devices (IoT), much lower latency. Aiming for standard approval in 2019, services possible 2020-21.
To complicate matters further, the ITU probably won’t designate any Pre-Release 16 version of 5G as 5G. It is also not certain that all of these flavors, or stages, of 5G will use higher spectrum bands or have the minimum channel bandwidth that was envisioned for 5G. For example, AT&T said that some of its ‘5G Evolution’ markets might include spectrum below 6 GHz.
This tells us that 5G has become as much a marketing game as it has a technology evolution game. We are likely to see some services marketed as ‘5G’ as early as this year. Some of these early 5G services might look like advances of the LTE roadmap, using channel aggregation and advanced antenna techniques. It might actually be difficult to distinguish between LTE Advanced Pro (marketed as 4.5G) and 5G. Interestingly, vendors have positioned this as 4.5G, but the operators for the time being are keeping all LTE enhancements under their current LTE branding (LTE Plus, LTE Advanced, etc.). Put in more simple terms, 4G Plus and 5G Minus might look very similar, from the standpoint of what’s delivered to the consumer. I’ve even seem the terms ‘enhanced mobile broadband’ used without calling it 5G or LTE Advanced Pro, further obfuscating the matter.
There is some precedent here. Remember that the first LTE services were introduced by Metro PCS in 2010, before Verizon and using only 5 MHz channels. Other operators called their services ‘4G’, even though they weren’t LTE: Sprint’s WiMAX, and HSPA+ services from AT&T and T-Mobile. And like we might see with ‘4.5G’, HSPA+ services outperformed LTE, in some markets and on some operators. It was very situational.
We are likely to see a similar playbook with 5G. You might have to come up with your own definition of what 5G really means. Is it based on a 5G New Radio? Does it have to use millimeter wave spectrum (i.e. 28 GHz or higher), or have a minimum channel bandwidth? What level of performance would signify a substantial enough improvement over 4G?
Remember, it was a good couple of years into LTE before we saw that it was a generational leap ahead of a typical 3G service – delivering 5-10x the throughput, IP-based (allowing for mobile VoIP services such as VoLTE), and some fairly significant leaps in capacity, among other enhancements.
So, consider this, in your own definition of 5G: You’ve got to tell a non-industry person at a cocktail party that 5G represents the ‘next generation’ of wireless. For me, that would mean, at a minimum:
- Throughput rates consistently above 100 Mbps, and closer to 500 Mbps
- Latency below 20 ms (good enough for some new use cases but not all)
- 10x capacity improvement on data volume
I’d accept a later stage version for some of the significant leaps in # of IoT devices supported, or the network architecture visions.
What’s your definition of 5G?
Mark Lowenstein, a leading industry analyst, consultant, and commentator, is Managing Director of Mobile Ecosystem. Click here to subscribe to his free Lens on Wireless monthly newsletter, or follow him on Twitter at @marklowenstein.