This year’s Mobile World Congress (MWC) Americas is, in many ways, a story of edges. Smartphone manufacturers such as Apple (which doesn’t need to actually exhibit to show its importance), Essential, and Samsung are clearly demonstrating that phones should have less edge, in the drive to create a single glass front. On the other hand, the networks that enable these devices will require a far greater edge presence to succeed in the future. Simply put, in the future world of 5G wireless, the simple goal of a “cloud” is not enough to ensure that the content the consumer wants will arrive quickly enough, and with a low enough latency, to be of value.
The bandwidth demands that we will be placing on the network, both as consumers and business users, is simply not possible without putting far more processing power closer to the point of consumption. Consider, for example, Verizon's virtual reality (VR) demonstration at MWC Americas, which included a live stream (in 4k) of a baseball game, as well as four other 4k streams on virtual TVs... as well as a smattering of VR games to keep you occupied in the seven inning stretch. That demo, running over a 5G test network, pulls a lot of bandwidth, and while the "last mile" may be addressed via wireless connectivity, the congestion starts to become an issue when you aggregate multiple consumer streams as you get deeper into the cloud, closer to the content source.
While the Verizon demo may seem like an extreme case today, it’s a safe bet that the bandwidth demands will grow incrementally over a relatively short period of time. The Automotive Edge Computing Consortium, which includes the likes of Toyota, Intel and NTT DoCoMo, estimates that the volume of data between just cars and the cloud will reach 10 exabytes per month (10 billion gigabytes) by 2025. Add in the VR and video viewing needs and suddenly there’s a bit of a problem with delivering the volume of data using the current model.
To address this, there’s a need to put far more of the processing power at the edge of the network, reducing the pressure on the core and helping to alleviate some of the bandwidth concerns. And the closer to the “edge” this processing can take place, the more efficient the network becomes, and the lower the latency will potentially be. However, this will drive up the cost of 5G implementation.
The squirrelly part of this is deciding just how close to the end consumer this edge needs to be, and how much processing powers is required. And as we all know from the constant need to upgrade our PC's and phones, the processing power we think we need today is often too little for tomorrow's needs. Again, the number 10 billion gigabytes comes to mind and with VR still very much in its infancy, making a solid guess on data consumption and the processing that these new applications will demand is no easy task.
Getting it wrong means either under-delivering and underwhelming the customer, which is pretty much the kiss of death for a mobile operator; while over-delivering means incurring unnecessary costs in a highly competitive, and increasingly margin focused, industry. Either way, we suspect implementing 5G with the future demands of consumers in mind is going to be a far more expensive and complicated affair than any previous network.