Future revenue opportunities seem very bright indeed for Internet of Things but to enable the business model and ensure workable profit margins, MNOs will need to look carefully at how they can save costs, writes Carsten Brinkschulte, the chief executive of Core Network Dynamics, in the second instalment of a two part article.

Based on new figures in the IoT Forecast Database Research published by analyst house Machina Research, the GSMA recently announced that mobile network operators are set to benefit from an estimated US$1.8 trillion Internet of Things (IoT) revenue opportunity by 2026 – boosted by the early deployment of commercial Low Power Wide Area Networks in licensed spectrum.

But to realise that opportunity, MNOs need to ensure that their networks are profitable. Currently, it costs a tier one telecom operator around US$8 to run the core network infrastructure for each smartphone; but shouldering such high costs is unsustainable for connecting smart water meters and the myriad of low-amp IoT devices that will flood the networks. MNOs will need to cut costs drastically to enable the IoT business model and dumping their data centre assets is an obvious way to make dramatic cost savings.

Staying nimble
In addition to the obvious financial savings, low latency is another attractive benefit offered by the public cloud.

Most latency occurs because of backhaul to the mobile core network but, as over 50% of traffic that goes to the mobile networks originates from the public cloud, it makes technical sense to put a virtualised, distributed EPC in the cloud, where the majority of the content is.

Otherwise, for private LTE networks and other remote communications use cases including emergency services, oil or mining and construction, as well as for industrial IoT (IIoT) use cases requiring ultra-low latency private networks, it makes sense to bring the distributed EPC on premise – or ultimately integrate the core network with the radio – and avoid backhaul altogether.

Take the example of a car manufacturer’s production line in which all the wireline network infrastructure has been replaced with a local wireless network. In this case, real-time applications such as controlling robots based on sensors and actuators is essential, but a centralised network architecture simply won’t cut it because network latency could be as slow as 100 milliseconds, which means that robots will not react quickly enough to information provided by external sensors.

By handling signaling locally and keeping local traffic local by having the factory robots communicating in real-time over a local LTE or 5G network, network latency can be radically reduced to a much more acceptable latency of 5ms or less.

The public cloud leap of faith
While previously unthinkable, it will be interesting to see which MNO breaks cover first and moves lock stock and barrel to the public cloud. The financial and technical pressures mean it’s not a matter of when but how soon it will be before a tier one operator strikes a deal with AWS, Google or Microsoft. And, when that MNO reports more cost efficiencies and increased profits, then competitors will follow.

Just as Phythagoras and Galileo challenged commonly held and cherished viewpoints around them, MNO CFOs are taking a long cool look at their financial spreadsheets, and will soon be ringing the changes.