Pivotal Commware (referred to as PC in the rest of the article) is a developer of software-defined antennas and radios designed to increase network speed, capacity, and spectral efficiency. Established in 2016, Pivotal Commware commercializes a breakthrough in electromagnetic physics, Holographic Beam Forming™ (HBF). The company’s wireless communication products utilize holographic beamforming technology that helps multiple concurrent transmissions using the same frequency without interference by creating broadband wireless networks, enabling wireless service providers to continuously reuse the same band of spectrum, at the same time, within a given spatial region.
One of the important data points used to define the project parameters and planning works, while rolling out infrastructure projects, is the significance of geographical and street features in a given area. These come in the form of various vectorized features such as building footprints and street furniture. The efficiency of obtaining these features and the level of accuracy are major factors that decide the success or failure of a telecom network rollout. Now, there were various issues which the planning teams at Pivotal Commware faced which were geospatial in nature.
Why was accessing accurate building footprints a challenge?
The team at PC had been using open source building footprints on a trial basis but they were running into multiple problems. One of the major issues that Pivotal Commware had to deal with in the former was data inaccuracy. Many polygons were loosely drawn polygons, positional accuracy was low and many building structures that had been recently constructed were not part of the database so their database was not updated with the latest on-ground changes. Inaccurate geospatial twins like OSM have far severe consequences on the quality of the network.
Take an example: open source building footprint databases do not capture structures like garages/sheds. That would mean none of these structures have been optimized for RF planning and hence the 5G network would not be as seamless in these areas leading to poor coverage and customer dissatisfaction. This was not acceptable to the highly quality sensitive team at PC.
The PC team that they would manually curate open source data. But manual correction of open source data was time consuming and costly and as more areas of interest get added to PC’s purview, the requirement for new and accurate building footprints shall increase even more dramatically. Pivotal Commware needed a scalable solution to solve this problem.
Pivotal Commware took a major step towards resolving these issues by engaging Attentive AI to provide accurate building footprints as a service. As a result of this, Pivotal Commware was able to easily request building footprints and other GIS features for any area of interest using high resolution aerial imagery and point cloud data. With the use of Attentive AI’s state-of-the-art geo-AI models, PC’s team data accuracy increased by over 50%. Not only that, it also minimized the turn-around time (for request completion) to a few days as opposed to a few months earlier. Additionally, the AI pipeline also allows them to capture on-ground changes by supplying fresh aerial data. The relevancy and value of data has thus increased drastically leading to the eventual end goal of seamless network rollout and higher customer satisfaction.
Are you fed up of inaccurate footprints too?
Till now Pivotal Commware has analyzed more than 20 million sq.m. of area using Attentive AI’s feature extraction engine. A detailed case study can also be found here. Read more on their success story and how they replaced an inaccurate building footprint data source with an accurate AI model based pipeline.