Monitoring the progress of 5G at the recent 5G Americas analyst summit, where carriers including AT&T, T-Mobile and Sprint engaged industry analyst on many 5G topics, I noted few observations. For context, the industry remains fixated on capacity: how to scale to support 11 GB in 2019 from ~4 GB monthly usage? This is fuelling the drive towards millimeter wave* communications (mmWave) which has become synonymous with 5G. While the access network remains the focus, I think this is taking away attention from other, more important, issues not the least of which is the use cases and applications that will generate revenue. This, in my opinion, could lead to future disappointments: the access network for 5G seems will be late in coming!
A few reasons lead me to this conclusion. First, LTE and its evolution has a very long life as has been well stated by the operators. We can safely say that LTE will be around 15 years from now if not even longer. This gives LTE deployments a 20-year lifecycle which is reasonable considering the success LTE has. The improvement in spectral efficiency that 5G will bring in sub 6 GHz bands is not substantial to warrant a rip-and-replace strategy – most sub 6 GHz spectrum will use LTE leaving little room for nascent 5G technology to take hold in that spectrum. Peak capacity enhancement is attained primarily though carrier aggregation which LTE supports, and through antenna-multiplicity which will have limited applicability in sub 6 GHz bands. LTE has been making good strides in reducing latency (only a few x’s from 5G requirements!). The improvements 5G brings are embedded deeper into the standard and focused on optimizing features and functions available in LTE. In other words, 5G will have no business case in sub 6 GHz bands in the near future.
Second, two critical aspects are underestimated in millimeter wave access: devices and transport. The first aspect would flush out the true use case of 5G: is it mobile or fixed? Indoor or outdoor deployments? [If fixed, even if anchored by a mobile network, why should it be called 5G?!!] Embedding mmWave technologies into mobile devices is no simple feat, especially considering the antenna requirements necessary to achieve range. Requirements for outdoor access are technically challenging. There are differing views on channel stability and performance which impacts service performance as well as the planning and deployment of mmWave sites an aspect that’s largely been ignored. Additionally, the ecosystem that’s required to drive down the cost of mmWave deployments will be missing. Markets expressing interest in mmWave technology – USA, Japan, and Korea – are not harmonized and alone are unlikely to generate sufficient scale to achieve necessary cost targets (could there be a case to roll 802.11ad into 5G?). Interestingly, vendors who see access infrastructure capex spending slowing down will be keen to have mmWave base station for carriers to test, but investment on silicon for devices will need much justification. This is a space to watch in the future.
In relations to the transport network, this issue transcends meeting the performance requirements of the access network to include design parameters and architecture of the transport network. Fiber will be required for mmWave transport which means access nodes will have limited deployments to where fiber exist (same issue as that of small cells). Designing for peak capacity, as is the practice today, will result in high asset underutilization. As a result, 5G will be expensive to plan and deploy in parts of the network that traditionally assume most capex: devices, access, and transport.
The evolution of 5G seen from the perspective of 2016 is one of a long road that will have many twist and turns and unexpected results. There are even greater issues, mainly focused on the business models of carriers that will need to adapt to 5G. All these elements make 5G unlike any of the previous technologies.
* Millimeter wave denotes frequencies over 30 GHz. However, the term became common to use for microwave spectrum in the 28 GHz as well, or even for any spectrum over 6 GHz.