The value of Network Infrastructure Management for mobile network carriers

From 2G to 5G, over the last few years the mobile network infrastructure has been populated by layers of devices with very different technological and performance characteristics. For this reason, modern network infrastructure managementservices have to operate over complex infrastructure, which is technologically heterogeneous, geographically extensive, with a widespread presence and tens of thousands of network nodes. 
But, at the same time, the revenue per bit transmitted by a mobile network is of increasingly higher value than that for fixed network equipment. In the mobile sector, indeed, users are willing to pay premium prices for quality services with the best user experience. In such a highly competitive scenario, a reduction in customer churntherefore becomes a key distinctive component. It is therefore understandable how strategically important the performance of the access network and the maintenance of high standards for the overall mobile customer experience is. 

A network outage or even a simple degradation in performance of the mobile infrastructure represents reputational damage for the operator, if not a financial loss. For this reason, all the Core Network systems of the carriers’ mobile networks have long been fully redundant, protected and appropriately oversized. 

The idea is to cope with unusual peaks of traffic during both extraordinary and scheduled events, such as a football match or a concert. In all cases of particular concentration of instances, the mobile network must be ready to respond adequately, just as for ordinary management. 

Until now, this activity has been managed in different ways. First of all, the sophisticated supervision systems of the Core Network equipment come into play. These systems are supervised by specialized personnel who is able to interpret the wide range of performance indicators, together with the metrics relating to traffic parameters and signaling of the control nodes and gateways that make up the Core Network.

Faults or exceeded thresholds considered to be critical are generally displayed through predefined alerts that indicate the critical status of the system. Moreover, the modulation of these alerts provides for specific intermediate stages involving checks on the parameters’ trends in order to allow prompt intervention before any anomaly should occur. 

This is an analysis task performed by operators including through the use of custom dashboards which are not included in the vendor’s supervisory tools. The dashboards offer unified monitoring of the high-level network parameters, values from equipment of different vendors which indicate the general state of health of the infrastructure. 

Summary graphs tracking these phenomena over time are created using the recorded values. Finally, logging of these graphs on a daily, weekly and monthly basis provides operators with agrid of benchmarks for the immediate identification of deviations of the values observed compared to those which are expected and plausible on a historical basis.

A traditional Network Infrastructure Management system, as designed, allows operators to react to faults but – with rare exceptions – not prevent them. In addition, the many alarms generated by these systems are likely to be ignored by operators, or it is simply not possible to monitor them all simultaneously in real time. In this increasingly complex scenario, the goal of mobile operators is to prevent the onset of network faults in a more functional manner.

The Core Networks, therefore, have traffic monitoring and analysis systems based on passive probes which gather the traffic intercepted by specific devices (packet brokers or taps) and send it to centralized systems. 

Moreover, some monitoring systems go as far as applying Business Intelligence processes in order to extract valuable data on network performance for use by CTOs through easy-to-use consoles. The final frontier in predictive monitoring concerns the insertion of Artificial Intelligence engines, especially in connection with the field of machine learning.

If this is the right direction, it should be noted that these modern Network Infrastructure Management systems require an experienced team capable of carrying out the initial training phase of the AI modules. 

Against this background, the carrier has the opportunity to rely on a competent partner who is ready to respond to these requirements. Lutech has the know-how and experience needed to install and develop these traffic analysis systems for the purposes of predictive maintenance. But they offer more than this. 

These standard systems, as previously mentioned, have functions that are not very versatile and are often still very basic and underdeveloped. A good alternative can come from the skills developed by system integrators in Machine Learning technologies. To overcome human limitations in the simultaneous supervision of hundreds of indicators and alarms from gateways and control nodes, automation tools can be used to monitor and interlink these variables in order to create patterns which are able to anticipate network failures or anomalies. 

Lutech has taken up the challenge in this area posed by a major Italian convergent network operator, with which it had previously worked on a number of intelligent automation projects for the optimization of network operations

Thanks to the problem-solving skills of Lutech’s Cognitive team, automatic “anomaly detection” mechanisms were created for the Core Network nodes and large amounts of data analyzed to set up predictive maintenance mechanisms, with great success.

The development and implementation of these tools was made possible by collaboration between Lutech’s development team and the operator’s team. A synergy necessary both to correctly interpret the variables observed and to direct Lutech towards the right analysis and prediction solutions.