5G: Italian telcos’ challenges amid smart planning and smart operations

We are witnessing the creation of a new society, hyper-connected and ultra-fast. In this context, CSPs are facing a new set of challenges, linked to the complexities of the new 5G technology, continuous market changes, the growing normative and regulatory pressure, and to a consumer base which, on the one hand, is getting increasingly attentive to the quality of the service provided and, on the other, to the monthly expenditure they have to incur.

In particular, the 5G network – which is a “network of networks” – has in itself an intrinsic complexity that is quite different from those seen so far. It can also enable new cross-market use cases, functional to support many investments with extremely different requirements: unprecedented speed, low latency and huge density of connections per km² (imagine the V2X, Vehicle-to-Everything scenarios).

To date, Artificial Intelligence has been primarily applied in its automation form, supporting human activities to perform them faster, more accurately and at a lower cost. 5G, due to its complexity, will be so dependent on AI in its broadest sense that it will not be able to reach its full capacity without exploiting AI’s potential for network planning, implementation, and management.

Analytics and AI have a wide array of applications in network planning, from traffic predictive modelling in each network segment to the further development of self-organizing networks (or SONs), up to advanced investment efficiency models aimed at maximizing KPIs that are more in line with a medium to long term strategy, mostly related to ROI and customer satisfaction goals.

More specifically, operators are aware that they need to shift their paradigm, moving from a network-centric perspective to an end-user-centric one, and that they also need to set investment priorities based on the service quality provided and on the user segments impacted. Therefore, in addition to cost and technological complexity, average revenues per cell and churn propensity become the criteria needed to choose between initiatives. We can thus talk about “next generation customer value management”, which is closely linked to network planning. This makes it necessary to start from the analysis of the areas to evolve, while identifying the customers with the highest value and highest churn risk. When assessing the investments to be prioritized, the indicators of service quality in a specific area must be crossed with these variables to maximize the investments’ impact on customers and to expand the network through strategically targeted initiatives.

At the same time, operations is another key area of AI application in the 5G deployment framework. New digital services require faster processing, while the use of AI is an obligatory step, since the number of transactions and the complexity of the service infrastructure are increasing.

In particular, there are three strategic areas where  AI in operations can be applied: Asset Monitoring & Predictive Maintenance, which exploits the high potential of IoT technologies for real-time monitoring of devices for predictive and proactive maintenance; Intelligent Automation@Operations, such as the gradual injection of automation and AI into service operation processes and the platforms it leverages on; and Augmented Reality for Field Service Assistance, i.e. the use of innovative augmented reality solutions capable of automating interactive and visual access to technical and operational information, facilitating communication between field technicians and remote operators as well.

These areas require a transition from “proactive” to “prescriptive” operational paradigms, based on a step-by-step approach that allows an initial small scale and short-time test of new models to be followed by a “use case factory” approach, aimed at the MVP development. The importance of data and data governance should not be overlooked: for many operators, the lack of a homogeneous and consistent information base represents one of the main obstacles to Artificial Intelligence’s adoption in all its potential. A “start small & think big” approach is key in this area, also suggesting to address initially-contained use cases to be optimized and scaled up gradually.

The article was drafted by Gea Smith.