If you believe all the hype around 5G – and there's plenty of it – it will deliver multi-gigabit, low latency broadband to mobile devices.  That “mobile device” may look like an automobile rather than an actual phone, but it's hard to say exactly what users are going to get when the smoke clears.  Regardless of the final speeds, 5G is going to force the entire network from edge to core to get an overhaul – otherwise what's the point of 5G in the first place?

First, let's set aside the “5G is needed for IoT” meme.  Providing an overlay network using low-power radios and devices for the Internet of Things is a separate project and specifications from the blazing speeds promised are being demonstrated in various bits and pieces by AT&T, Sprint, T-Mobile, and Verizon along with Ericsson, Nokia, and various Asian network equipment manufacturers.  At some point, someone in the market department is going to split 5G into 5G-IoT and 5G-Fast, where 5G-Fast is the heir to LTE after LTE is milked for every single megabit of speed possible through existing licensed spectrum, plus whatever it can steal, er, borrow from Wi-Fi bands.

The world of 5G-Fast promises translate to using millimeter-wave technology – basically line-of-sight – which may or may not be licensed, depending on the country and regulation in place – with latency on the order of under a millisecond or less.  Compare that to 4G LTE with operational latency on real world networks in the ballpark of 50 to 60 milliseconds or so.  Available speeds on an LTE network depending on what version of tech, with download speeds possible in the 400 to 450 Mbps range on some carriers, such as T-Mobile.

LTE has plenty of room for speed growth.  Telstra, Ericsson, and Qualcomm pushed the state of the art, demonstrating download speeds of 979 Mbps with upload speeds of 129 Mbps during recent test results announced in September.  It's a given that hardware vendors and carriers want to squeeze out up to 1 Gbps using aggregation of various radio frequency bands, multiple antennas, and advanced data encoding schemes already in use on the wired side of the world.

 I figure LTE upgrades now and in the pipeline will keep the wireless industry busy until 2020 – the year we should see formal standardization of 5G. Standardization is important because it means interoperability.  Carriers don’t want to be locked into a single vendor, no matter how fast or good the technology.

Between now and 2020, carriers have four separate problems to address in preparing for 5G-Fast.  The 5G-Fast ready network will need faster network speeds, more cells, more fiber, and lower network latency.  Supporting multi-gigabit wireless speeds at the edge of the network will require 100 Gbps and faster speeds on backhaul with terabit and faster at the core.  “Maxing up” LTE speeds to the 1 Gbps range will lay the groundwork for supporting 5G class multi-gigabit speeds at the edge.

Moving into multi-gigabit speeds also means shorter range, line of sight or near line of sight cell distribution at the edge, given the use of millimeter wave radio frequencies.   If you thought the discussion of small, nano, and picocells for LTE deployments was fun, you will love 5G hardware and hate the planning necessary for optimizing 5G-Fast cell buildouts. In-building coverage will get more interesting and complex.

To support all of these new cells will require more fiber in more places while a “traditional” cell tower will end up being a 100 Gbps to 1 Tbps hub.  Verizon’s long hiatus on fiber deployments was broken in part with the company’s need to deploy more fiber in Boston for supporting LTE.   

Lowering latency will likely be the biggest challenge for service providers. Existing infrastructure is built around LTE 50 millisecond latency, not 5G-Fast’s promise of 1 millisecond. Current network designs will need to be modified in order to support lower latency, as well as take into account additional back-end latencies added with more cells to support 5G line-of-sight.  If the core network can’t keep up with 5G’s lower latencies, it’s going to be an issue somewhere down the road.