"As the task of managing networks becomes more complex," says Vodafone's head of Network Strategy and Architecture, Santiago Tenorio, "our engineers can use machine learning to release themselves from routine tasks that are very time consuming to focus their effort on more critical and strategic projects."
Tenorio posted details on this development in an official Vodafone blog post.
He reports that, "Vodafone Germany and Huawei recently trialled machine learning in a Centralised Self-Organising Network (C-SON), to identify the optimal settings to deliver voice over LTE services across 450 mobile cells chosen at random".
"The algorithm completed the task in four hours. The same task would have taken an engineer about 2.5 months to do manually."
"The technology works by monitoring network traffic trends. This monitoring complies with Vodafone’s strict commitment to customer privacy and no data is involved that could be used to identify individuals."
With network optimisation speed increased by a whopping 45,000%, Tenorio says the programme "predicts future network traffic behaviour based on data processing and pattern recognition".
He continues, stating that to date, "all high traffic predictions have been correct. The predictions enable the network to self-configure itself automatically to balance the traffic load among neighbouring cells and improve the customer experience.
"Initial results confirmed an average 6% improvement in the mobile download speed and lower interference at the cell sites (the cause of dropped calls, problems connecting and higher device battery drain)."
Tenorio notes that: "Vodafone customers could experience significant benefits from the use of machine learning in our networks".
"For instance, the network could identify if there is high traffic at a mobile cell site every Thursday at 8pm — perhaps generated by weekly concerts at a popular music venue — and automatically increase the cell’s capacity before people arrive, returning to normal after they go home.
"Customers would benefit from the uninterrupted ability to call, message or share videos and photos on social media throughout the night."
He says "further machine learning trials [are] planned and he expects to "begin utilising the technology in our commercial networks during the 2018/19 financial year".