There is massive disruption in the telecoms space, driven in the main part by the shift from proprietary hardware to commoditised hardware, and by the virtualisation of network functions in software.
We are seeing this in the core network, the radio network, and at the network edge, as compute power spreads further out to the margins. The theory is plain: networks will become more flexible, controlled in software; their operators will be able to improve efficiencies and reduce costs.
The pace of innovation will jump, and the market will, at last, be competitive with over-the-top (OTT) internet providers.
But this idea it gets simpler with software-defined networking (SDN) and network-functions virtualisation (NFV) is wrongheaded – or at least, it misses an intermediate step. Because it isn’t getting any simpler; it is getting harder, and more complicated – network operators are left with virtualised chaos, instead.
Order must come out of this chaos, or hopes that the telecoms sector will reinvent itself with network technologies like 5G and network architectures like the ‘internet of things’ (IoT) will be dashed.
The telecoms industry is presented, by luck and design, with an opportunity to be an agent of industrial change, as well as its subject. It holds trump cards in this transformation game; to an extent, it has some control of the hand.
The fourth industrial revolution, as it gets called, is a chance for telecoms providers to target new verticals, like cities and manufacturing, with ultra-reliable, low latency communications – made bespoke for industry by features like network slicing.
But the rising chaos is threatening to stop that. The number of parameters on a base station has gone from 500 in the 2G era, to 1,500 with 3G, and 3,500 today, way before 5G rolls out. Automation tools exist to tune these settings, but more needs to be done if wireless networks are to be trusted by industry to support their most critical systems and services.
Which is where AI comes in. If operators can manage the chaos, they can get out of their negative spiral – and move from commoditised services to value-added services. AI is like the skeleton key to change for the telecoms industry, whichever door it takes.
More than this, the telecoms industry’s take on AI is key for other industries as well.
At the moment, even the most celebrated cloud-native brands – like Netflix and Spotify – survive on a best-effort service, dictated by the quality of the signal to the end user’s device. In the future, with ultra reliable 5G, this best-effort delivery will not do.
It will not be acceptable for driverless cars, it will not be acceptable for high-end manufacturing, it will not be acceptable for any kind of high security in any kind of industrial networks. It will not be acceptable for any of the new industries the telecoms market wants to serve.
The quality of experience has to be better. It has to be completely reliable. Networks have to move in response to, and anticipation of, changes in their infrastructure in order to guarantee services. The only way to do that is with AI.
