Using AI to Increase the Accuracy and Efficiency of Automatic Fiber Network Planning Activities

Telecommunications Firm

Using AI to Increase the Accuracy and Efficiency of Automatic Fiber Network Planning Activities

Through the correction of map distortions and object detection in hand-drawn maps

The Challenge

Fiber network often lacks extensive digitization of location maps. On top of this, the digitization process of missing information is unmanageable by human workers. A telecommunications firm called for our help in overcoming these problems. The company in question had limited automated planning capability for fixed networks due to insufficient data quality, and there were distortions in its hand-drawn maps. It had also incompletely digitized network objects like pipes and their configurations. Thankfully, we knew what we could do to help.  

The end result? Improved planning accuracy and efficiency of the automated fixed network rollout through map data enrichment. 

The approach

Firstly, we divided the company’s challenges into three subproblems: detection, matching, and correction. We then accurately identified buildings in location maps using a fuzzy classifier, taking into account contour hierarchies.  
Next, we used tools such as a geometrical moment of inertia, relative positions on the map, and meta information (such as street names and house numbers) to match detected buildings with reference data.  
Finally, we corrected distortions within location maps based on the center point of the area and interpolation between building pairs. We also detected tube symbols and related textual information on the maps. 

The result

Our help in this project brought about a successful algorithm for the automatic detection of buildings in hand-drawn maps for the telecoms company. The firm also benefited from having matching algorithms for detected and reference-building objects.  
The correction algorithm we helped build, which was based on detected buildings, helped to correct distortions within the location maps. 

Let’s create sustainable change together.