DT searches for fibre efficiency with Fraunhofer AI

Sometimes there are stories which come along and prove stereotypes can be true. In the search for efficiency, Deutsche Telekom is turning to artificial intelligence to help with its fibre rollout plans.

Partnering Fraunhofer IPM, DT has unveiled a pilot project where artificial intelligence will look at images and information gathered by a measurement vehicle, before deciding what the best way to dig trenches and lay fibre will be. The pilot will take place in Bornheim, near Bonn.

“The shortest route to the customer is not always the most economical,” said Walter Goldenits, Head of Technology at Telekom Deutschland. “By using artificial intelligence in the planning phase we can speed up our fiber-optic roll-out. This enables us to offer our customers broadband lines faster and, above all, more efficiently.”

The measurement vehicle is equipped with 360° cameras and laser scanners, and will collect roughly 5 GB of surface data per kilometre. Depending on the terrain, the vehicle can cover 50-80 kilometres per day, collecting information such as the location of trees, the type of ground which will need to be dug up and if there is street furniture which needs to be accounted for. When deploying new infrastructure, engineers have to ensure the environment is returned to the same condition as before; various scenarios can have different impacts. Sometimes it could be more time and cost efficient to go the long way around.

“Such huge amounts of data are both a blessing and a curse,” said Dr. Alexander Reiterer, project lead at the Fraunhofer IPM. “We need as many details as possible. At the same time, the whole endeavour is only efficient if you can avoid laboriously combing through the data to find the information you need. For the planning process to be efficient the evaluation of these enormous amounts of data must be automated.”

The neural network used for this recognizes a total of approximately 30 different categories through deep learning algorithms, including trees, street lights, asphalt and cobblestones. The applications can even identify whether the pavements feature large pavement slabs or small cobblestones, if the trees deciduous or coniferous, or whether the trees roots will impede the engineers during the project. Once all these factors have been taken into account, the existing infrastructure is assessing before decisions are made and an optimal route planned for the new fibre.

Such a project will capture the attention of many around the world. The rollout of fibre has been staggered and slow to date partly due to the expense. Of course, the raw materials are expensive, though digging trenches and laying the cable is an laborious, costly and slow process. Telcos will of course be looking for new ideas to keep the cost down, though governments will also be peering across as they increasingly demand faster deployments.

Back in April, the FTTH Council Europe unveiled research which demonstrated quickly some countries were progressing with fibre rollout, though it also shed light on how woefully terrible others are doing. Latvia led the way with 50.6% household penetration, though Ireland’s was down at 1.7%. The research did not include the UK, though Ofcom’s estimates put FTTH penetration down at 3%; definitely in the woefully poor category.