In this thought piece, NSSN Data Theme Leader, Ron Johan, gives his perspective on improving the efficiency of telecommunication networks using big data.

Originally telecommunications networks were quite simple, comprising a copper cable that carried a DC voltage that was modulated by speech signals. While these were simple and robust, if something did go wrong a technician was required to analyse the entire end-to-end circuit to find the fault. The system provided very little in the way of diagnostics to support fault localisation.

The situation began to change with the introduction of digital communications during the 80s. Now both the exchange end (LT - line termination) and the remote end (NT – network termination) were able to generate diagnostic data. Furthermore, the remote end was able to use spare capacity in the digital bearer to deliver the diagnostic information generated by the system to the exchange and from there to a network management centre. While this was not ideal, it was significantly better than the old analogue implementation as it supported a measure of remote fault-finding.

Modern telecommunications networks, such as those used to support smartphones and landline based broadband delivery are quite complex. Both ends of the system are now capable of generating vast amounts of data detailing the performance of the system.

However, due both to cost pressures and the availability of very small surface mounted components, currently, a single interface card located at the exchange now supports dozens of customer broadband services. That is, if one interface card fails in the exchange equipment, multiple customer services drop off the air.

While a single outage brought about through the failure of a customer interface card is relatively straightforward to identify and rectify, intermittent faults that involve multiple services can be quite challenging to address as the fault may be brought about by an unanticipated interaction between different interface cards in the exchange equipment. Often it takes the customer noticing that their broadband service is unreliable and complaining to their provider. The intermittent nature of the fault generally results in a high level of customer frustration with the service provider.

Tracking down these types of faults requires the collection of large amounts of performance data from the exchange equipment as well as from the customer premises equipment. The data obtained from the exchange has to be cross-correlated with the customer’s equipment in order to identify a pattern that links the unstable broadband connection with other broadband connections.

This work needs to be carried out by a skilled technician with a good understanding of the various statistical tools that are available. In a large and complex broadband network, there could be hundreds of such faults. The whole process is inefficient and cumbersome.

There is a clear role for the deployment of advances in Artificial Intelligence (AI) to revolutionise service delivery by internet providers, saving money for telcos, and improving customer experience.

A better approach would be to develop a tool that uses AI to identify fault correlation between customer services. The tool, running continuously across the telco network would identify faults, only triggering human intervention when a fault is detected. The result? Cost savings for telcos and improved experience and reliability for customers.

NBN Co and Australian telcos are investing millions each year on R&D to improve and enhance their service offering. It is critical that AI not be excluded from this investment. While the initial outlay in developing AI tools for telecommunications may be high, the benefits elucidated above are clearly evident and will return dividend year-on-year.

There is no need to look abroad for the kind of expertise to lead this AI revolution. With eleven world-class AI research centres across the NSW Smart Sensing Network, access to the brightest minds in the field are right here in NSW and the ACT. A smart investment now means a more connected future.

Ron is the NSSN Data Theme Leader, responsible for engaging with customers on advanced data analytics and IoT solutions.

Prior to joining the NSSN, Ron has spent many years in the telecommunications industry where he was responsible for Access Products. Ron went on to start up a company that develops a range of IoT solutions which it supplies to government, telcos and banks, as well as to major fire and security companies, to support the migration of their services from legacy telecommunications networks on to modern broadband and cellular technologies.

Find out more about Ron here.

Using big data to fault-locate telecommunication networks