Putting Broadband Coverage Gaps on the Map

Sypnosis

KPI

Successful implementation of a system powered with computer vision-based location regression accurately estimates the distance of addresses to close internet towers.

Project length:

Three-Week Intensive Bootcamp Training

Employees trained:

16

At a Glance

Broadband access is a necessity in our society, only further emphasized by the global pandemic. In order to get needed internet service to all citizens, gaps in internet service to rural areas of Canada had been determined with remote satellite imagery and calculations between physical addresses and towers.

However, researchers at the Communications Research Centre of the Canadian government found that their calculations were prone to error, often leading technicians and engineers to waste time and resources by going to inaccurate (and often remote) locations, unable to start work because the data was off. Data Society, through a three-week intensive, successfully led researchers to fix their existing tools for assessing gaps in broadband access ahead of the unanticipated pandemic.

Client Profile

A division of the Government of Canada’s Department of Innovation, Science and Economic Development, the Spectrum management and telecommunications team provide access to the radio frequency spectrum by issuing authorizations for its use, minimizing interference, securing Canada’s access to it through international negotiations, and ensuring its safe and efficient use.

The Challenge

To provide reliable internet service to rural communities of Canada, internet service providers need to be able to accurately estimate the distance to the closest tower. At the time, the Government of Canada’s system allowed them to estimate distances between physical addresses and towers via satellite images, but they were significantly inaccurate. Engineers weren’t aware of this until they arrived on-site, resulting in higher project costs and delaying setup of broadband service in communities, or prohibiting installation altogether.

The Communications Research Centre (CRC), a branch of the Spectrum and Telecommunication Sector of the Canadian government, issued an RFP for an intensive data science training course for 16 of its research staff to accelerate their ability to meet the increasing demand for data science solutions to spectrum research problems.

The Solution

After winning the RFP, Data Society went on-site to Ontario to offer a three-week course for the CRC. In order to convey the full breadth of information to be covered (140 hours of material), the Division halted all other work to focus only on the training – Monday to Friday, eight hours each day.

Approach

The training was optimized for immediate applicability, designed around the existing datasets, code environment, and machines the CRC was using for the spectrum program.

Mentorship Structures and a Quick COVID-19 pivot

A parallel objective of the training was achieving greater synergy across teams. The goal was to understand where the effective impact was being made and to create synergies where there were currently silos. 

The core team of five experienced researchers aspired to catch up with the latest skills and techniques, while the rest of the team consisted of more junior researchers needing to hone skills. During the first two weeks, Data Society worked with the core team and collaborated with junior researchers to keep everyone up to speed, with additional self-guided resources provided to ensure the pace of learning could be flexible to the individual.

The third week of training was specifically created for the core team, allowing them to develop deeper skills with hands-on workshops with a skilled trainer, which proved to be highly effective as students approached their capstone project to map broadband access coverage.

In the spring of 2020, with the arrival of COVID-19, it soon became clear that in-person classes wouldn’t be possible. Data Society redesigned the training to fit a completely virtual format, and the core team of researchers received the “hands-on” training to solve their primary challenge.

The Results

Mapping Coverage in Rural Canada

Researchers were able to work through a large amount of data and research and improve their existing tool for estimating the distance between addresses and towers.

Now, the CRC can more effectively and accurately leverage data and satellite imagery to map broadband access coverage and precisely determine where further effort is required to bring broadband to rural areas of the country.