There is a Better Way to Measure Broadband Deployment
August 29, 2019 | by Andrew Regitsky
The deployment maps broadband providers are required to generate vastly overstate actual broadband deployment. Thus, USTelecom and a bunch of other providers, including ITTA, AT&T, WISPA, Chariton Valley, CenturyLink, Consolidated, Frontier, Riverstreet, TDS, Verizon and Windstream, tried to find a better way. Instead of relying on census blocks, the consortium began a “Broadband Mapping Initiative.” This was highlighted by a pilot program in which every location in Missouri and Virginia was aggregated to create a “Broadband Serviceable Location Fabric” to identify specific locations that need broadband access. Here’s how it worked:
- Multiple sources of address, building, and parcel data were used to develop and validate a comprehensive database of all broadband serviceable locations in the two pilot states.
- A vendor conformed address formats, remove duplicates, and using a geo-referencing tool assign a unique latitude and longitude to the actual building where broadband service is most likely to be installed.
- Customer address lists provided by participating companies augmented the validation process and were automatically indexed to the final database to facilitate accurate broadband availability reporting. Different methods for reporting service availability were tested.
- The pilot also developed and tested a mediated crowdsourcing platform that will enable consumers to submit information to improve the accuracy of the database.
The pilot was expected to take four to six months. Five months later it was completed, and the results made public on August 20, 2019 in an ex parte letter from USTelecom to the FCC. According to the broadband association, 38 percent of additional rural locations are unserved in census blocks that would have been reported as having available broadband service using the FCC Form 477 generated maps. Here are more key findings:
The Pilot was a Success: Using state of the art technology and a combination of public and commercial datasets, it is now possible to identify and precisely locate virtually every structure in a geographic area that is capable of receiving broadband. Developing the Fabric for two states shows it is possible to do so for the entire country.
Pinpointing Service Availability: Creating the Fabric revealed that in just two states over 450,000 homes and businesses exist that are counted as “served” under current 477 reporting that are not receiving service from participating providers. While not every broadband provider chose to participate in this Pilot – so the actual number of unserved may be lower – that still leaves the potential for substantial misrepresentations about service availability.
The Counts Count: We measured broadband availability by locations in a census block. The Fabric revealed that 48% of the location counts in rural census blocks are different from current estimates, in many cases significantly different.
Timely and Cost Effective: A nationally developed dataset of all broadband serviceable locations consistent with the approach demonstrated in the Pilot can be achieved in 12-15 months. The cost to do so will vary depending on the mix of open source or proprietary data sources, but a national Fabric could potentially be developed for between $8.5-$11 million in upfront costs and $3-4 million in annual updates.
Location, Location, Location: Broadband availability is about connections, but providers must know exactly where a structure is in order to provide that link. The presumed geocoded location for 61% of rural homes and businesses are off by over 7.6m (25 feet) and 25% are off by over 100m (328 feet) – more than a football field! This distance can significantly impact the cost to deploy to an unserved location – making or breaking a decision to deploy for a provider.
Reporting Enhanced: Regardless of format (shapefile, propagation map, address, etc.) the quality and validity of reporting is driven by the quality of the underlying data on which the report is overlaid. (USTelecom, Docket 19-195, filed August 20, 2019, at Executive Summary).
USTelecom estimated that if the FCC adopts the methodology used in Missouri and Virginia for the entire country, it would take 18 to 24 months.
The results here will become part of the FCC’s attempt to obtain more accurate broadband data in ongoing Docket 19-195. In that proceeding, the Commission initiated a new data collection, the Digital Opportunity Data Collection, to gather geospatial broadband service. All broadband providers must submit granular maps of the areas where they have broadband-capable networks and make service available. Specifically:
Service providers—who are uniquely situated to know where their own networks are deployed—must determine in the first instance the availability of broadband in their service areas, taking into account their individual circumstances and their on-the-ground knowledge and experience. At the same time, to complement this granular broadband availability data, we adopt a process to begin collecting public input, sometimes known as “crowdsourcing,” on the accuracy of service providers’ broadband deployment data. Through this new tool, State, local, and Tribal governmental entities and members of the public will be able to submit fixed broadband availability data, leveraging their experience concerning service availability. (Docket 19-195, Report and Order, released August 6, 2019, at para. 3).
Unfortunately, in this proceeding, providers are permitted to use their own technical standards to determine where broadband is available. It would make more sense for the Commission to initially analyze the USTelecom results to determine if they are viable nationally, so all providers could use the same standards and procedures. As it stands now, all providers must conduct their own time-consuming mapping, only to potentially find that the USTelecom methodology is adopted. What a waste of company resources!