X
Tech

NBN launches Tech Lab to collect data and resolve connection issues

NBN will use its new Tech Lab to collate and learn from both customer experience surveys and current fault-reporting data using open-source machine-learning and big data software.
Written by Corinne Reichert, Contributor

Australia's National Broadband Network (NBN) company has announced its new Tech Lab, which will utilise big data, graph technology, and machine-learning capabilities to help solve and map end-user connectivity issues.

The NBN Tech Lab, once complete, will collect and collate data on user experience via voluntary surveys to be handed out to end users, enabling NBN to detect patterns in problems and user preferences.

It will also integrate all of NBN's current fault-reporting techniques, including the information generated by truck rolls.

"While for the majority, the installation experience is positive, when faults do occur, NBN's Tech Lab will help the team determine whether a fault can be dealt with remotely and immediately, or whether a field technician needs to visit an end-user home to resolve the fault," NBN explained on Thursday morning.

"The Tech Lab will also help NBN better understand the key factors that drive dissatisfaction and address them so people have a better experience."

The only data to be used by the Tech Lab would be publicly available or voluntarily submitted information, with the company now investigating the best platform to suit its needs.

As a result, NBN is looking into potentially using open-source technologies Apache Spark, Kafka, Flume, Cassandra, and JanusGraph for the project, alongside partner technologies such as Amazon Web Services (AWS) S3 storage, RStudio, H2O.ai, and ArangoDB.

"Our Tech Lab sees us utilising existing capability to solve a complex problem, and will help provide us with crucial insights about the way people are using the NBN," chief systems engineering officer John McInerney said.

"Developing these insights will help enrich the customer experience of services over the NBN access network and make our systems and processes more agile by synthesising massive data sets.

"Once the investigation and implementation of the Tech Lab research is complete, we could, for example, easily identify trends that occur in a failed activation in order to pre-empt problems before arriving at a house."

McInerney said NBN expects to see "significant improvements" in customer experience by simplifying and speeding up the fault detection and resolution process.

According to NBN, its Tech Lab became necessary as the rollout gained pace to connect around 45,000 premises per week, with the rollout expected to be 97 percent complete by June 2019.

NBN has expressed an increased focus on end-user experience of late. Earlier this week it announced trials of a new diagnostic tool aimed at remotely finding out whether a premises has copper wiring faults.

The copper cabling running to and from telephone sockets inside homes will be tested, as NBN said this can have a "very real impact" on broadband service quality as demonstrated earlier in the year by an internal study involving 800 fibre-to-the-node (FttN) premises.

"In coming weeks, we will begin a trial of a new diagnostic tool that we hope will quickly and accurately detect premises that may be suffering from speed issues related to in-home wiring faults," NBN acting CTO Carolyn Phiddian said on Wednesday, after the test documents [PDF] were published on Tuesday.

Cases of wiring faults could also affect those connected by fibre-to-the-building (FttB) and fibre-to-the-curb (FttC) network technologies, with NBN saying solutions include moving modems to the first socket, closing off unused phone outlets, installing a central splitter, and re-cabling poor wiring -- which could result in a download speed increase of around 30Mbps to 46Mbps.

Over the weekend, Communications Minister Mitch Fifield also said connections to the NBN are done right around 90 percent of the time now, with both NBN and the Australian government working towards improving end-user experiences.

Editorial standards