By Zita Goldman

5G SA and the IoT: is the arranged marriage between these two emerging technologies made in heaven?

Could 5G and edge computing eventually become the dominant technologies for IoT?

 

Misnomers are rife among cellular network standards. 4G LTE (Long Term Evolution), although it sounded fancier than plain 4G, was only an improved 3G. LTE-A, a more advanced version with boosted bandwidth, was the true 4G. And the latest, highest-performing LTE-A-Pro sports-standards, such as a 2Gbps speed and two-millisecond latency, are on a par with entry-level 5G.

 

Low latency – or a near-zero time between request and response – is expected to be the name of the game in the dawning 5G era, in the same way lag was the hallmark of pre-4G networks. However, reviewers of 5G-enabled smartphones testing 5G networks are often underwhelmed by both speed and latency. Why is that?

 

Well, 5G comes in three different shapes and sizes depending on the frequency it’s deployed on. The lowest band is reserved for the foot soldiers, and above 20 GHz you have the high-frequency superheroes who whizz at almost lightning speed in real time, the kind of image that the term conjures up in many of us. Mid-band with the best speed-penetration trade-offs is in between.

 

But 5G can’t reach its full potential in its non-standalone (NSA) form. Today, both the superhero and the foot soldier can reach their maximum speeds only between devices and 5G cell stations. From that point on, they piggyback the LTE infrastructure, which, even if fast, can add as many as 60 milliseconds to the trip, especially if it needs to get a response from a distant cloud.

 

This may sound too nuanced. And indeed, even Formula1 racers, the champions of human reaction speed, are in the hundred-millisecond range, with the majority of us taking about three times longer.

 

However, enhanced mobile broadband (eMBB) is only one of the three use-cases of fifth generation cellular networks. The other two – massive machine-type communication and ultra-reliable low latency communication (URLLC), meanwhile – are closely, although not exclusively, linked to the IoT, which involves a much faster M2M (machine-to-machine) type of communication.

 

Device density and zero latency – the IoT-related use cases for 5G

 

Like it or not, sensors are becoming ubiquitous. They are tagged onto water meters, white goods, streetlights, electric bikes and even cows. With sensors becoming ever-denser, they will soon be able to overwhelm 4G’s capacity of supporting 2,000 connected endpoints per square metre – while 5G can deliver a density of one to 10 million devices within the same area.

 

Abundant as they may be, latency for these kinds of sensors is in the majority of cases is not a priority at all. They transmit small packets of data periodically, and thanks to the latest technological advancements, they can even power out between transmissions.

 

Where latency – or rather the lack of it – is pivotal are nascent technologies such as autonomous cars, drones, telemedicine and augmented reality solutions. These are indeed areas where real and virtual time need to be in synch with milli- or even nanosecond accuracy.

 

Where is the edge and what does it bring to the table?

 

Although for the last five years there’s been a lot of talk about transferring compute and storage to hyperscale clouds, the State of the Edge Report 2020 has identified an underlying trend of getting both functions closer to the end-user. The departure point – or as the report labels it, Act I of the play – was the connected internet of the 1990s, a network of networks, where any device could be connected to any server. Regional data centres and Content Distribution Networks (CDNs) enabling video streaming and instant page loads form Act II, the first stage of bringing internet infrastructure closer to users. Act III, the way the report envisages it, is about the shift to the edge – where as many of computing processes happen at the point of request as possible.

 

This is the kind of rearchitecting, the report maintains, that can support new IoT use cases. If computing happens on the edge, the length of round trips between devices and servers will be dramatically reduced. There will be fewer hops through gateways, and interconnections and outdated back-end network architecture can be bypassed. The petabytes of information collected by trillions of sensors can be whittled down and processed by edge computing, which will decrease the load and therefore boost the speed and the uptime of the internet. All these changes will impact latency rather favourably.

 

The edge itself is currently hard to clearly define. It can be in devices such as EV charging stations, security cameras, smartphones, sensors, industrial robots or game consoles. It can also be placed at different points along networks: base stations, aggregation hubs or regional data centres. The State of the Edge Report predicts that about two-thirds of deployments will be in the infrastructure and one-third on the device edge by 2028.

 

Densification: the common enabler for 5G and the IoT

 

There are already a number of widely-adopted non-cellular network protocols for the IoT – just within LP WAN, the long-range, low bit-rate, low power category suited to remote, inaccessible sensor applications – such as LoRa, sigfox and Weightless. There are two things, however, that may work in favour of cellular IoT adoption.

 

One has to do with the Achilles heel of the millimetre-wave 5G superhero: low penetration. It has difficulties transmitting through obstructions such as walls, windows or even leaves. Therefore, to achieve the 100GHz bandwidth and 1ms latency target of the protocol, the existing network of base stations needs to be augmented with a dense mesh of small cell stations at a 0.6 mile distance from each other. These small cell stations look like good candidates for doubling as edge computing nodes that the IoT is after. 

 

In dense urban areas these small cell stations can comfortably perch on street furniture, but they may be too ungainly and costly to provide coverage for wide rural areas. One emerging way for the IoT to circumvent the lack of millimetre-wave 5G coverage is to set up a private industrial network of outdoor and indoor small cell stations. German industrial IoT pioneer Bosch, for example, already started the first 5G trial in its Worcester factory early last year, applying for localised spectrum licences in Germany last November.

 

Companies in the industrial, enterprise or smart city IoT arena who can afford to follow in Bosch’s footsteps will be the next to take full advantage of the complete 5G offering. Millimetre-wave 5G will fulfil on the promise of ultra-reliability by allowing the creation of multiple virtual networks atop a shared physical infrastructure. This capability, called slicing, will eventually do away with bandwidth sharing and the unpredictability of existing networks and has the long-term potential to turn real-time analytics, remote surgery and autonomous cars into reality. The same network will be able to run the 5G trump card, NB-IoT – the cellular answer to low bit-rate, low-power wide-area networks – thus providing interoperability between zero latency and low-power massive MTC use cases in the long run.

 

Meanwhile, the majority of companies will have to wait a few more years until the full roll-out of 5G SA – the pure, standalone 5G network.

Related articles

What's next?

Get our latest features in your inbox

Join our community of business leaders