Why the FCC Should Adopt CalSPEED, Part 1

The FCC's 2016 Broadband Progress Report quotes a number of data sources to answer what I’ll call the Big Policy Question posed in section 706 of the Telecommunications Act of 1996 – namely, whether “advanced telecommunications capability” is being deployed to all Americans in a reasonable and timely fashion. Unfortunately, the FCC's mobile data falls short in helping answer that question. I’ll demonstrate how and show why the FCC should adopt the CPUC’s mobile testing methodology.

First, though, we need to address the following questions regarding the data:

1.      What is the speed that consumers are getting compared to what is advertised?

2.      Is speed the only consideration?

3.      Should the basis of analysis be households or people?

Question 1 - the FCC has shown through its Measuring Broadband America reports that actual speeds are pretty close if not higher than advertised speeds for wireline technologies. For mobile, the answer is not so simple. However, the CPUC’s mobile testing program has produced mountains of data on availability and variability for California, and we find that actual speeds and performance vary greatly depending on whether you are in urban, rural, or tribal areas.

Question 2 - the FCC has considered other metrics such as latency and variability, but for now has excluded them from the definition of “advanced telecommunications capability” because of lack of data. Nevertheless, data from the CPUC’s mobile testing program has allowed us to create “heat maps” predicting coverage for varying service levels – E-mail, Voice over IP, streaming video, and video conferencing.

Question 3 is rhetorical. Mobile networks serve people. Sometimes, people are inside their homes, and other times they are on the move. Hence, mobile deployment and availability needs to focus on where people go, not where they are when they sleep at night.

So, how effective is the data used by the FCC in answering the Big Policy Question? In Part 2, I will summarize the pros and cons of the CPUC’s mobile field test data as well as those for three mobile data sets referenced by the FCC on their web site and in the 2016 Broadband Progress Report.

CalSPEED now available on iPhone

We are ecstatic to finally provide an iPhone version of the popular CalSPEED mobile testing app. This is the first full beta release of the app, and we are already working on improvements to the user interface and the mapping feature. Some users told us that the "Unsatisfactory/Satisfactory" readout at the top of the results screen is misleading, because it applies only to the Mean Opinion Score, not for the entire test. Nevertheless, nearly half of the 784 CalSPEED tests since the soft launch of the iPhone version last December have been run on iPhones. 

8th Round of Field Testing Completed

Last Friday, which was the final day of testing for this round, one of our testers celebrated with an ice cream and some trace route tests. Congratulations to the staff at CSU Chico and CSU Northridge for completing this round in a timely fashion despite the inclement weather.

Here are some more photos from this round:

More On Video Streaming: FreeBee, Sponsored Data, Etc.

Well, it turns out T-Mobile isn't the only provider offering data-charge-exempt streaming video for select content providers. Verizon and AT&T offer services where content providers pay for the data charges rather than the end user. Verizon's service is called "FreeBee Data," and AT&T's is creatively called "AT&T Sponsored Data." Sprint has yet to offer a similar service on its own network, but they do offer data-free music streaming for both Virgin and Boost MVNOs. So, with that, here are preliminary (refer to previous T-Mobile "Binge On, Binge Off" blog for why this is preliminary) estimated streaming video coverage maps for the other 3 major providers in California.

Verizon Wireless

 

 AT&T Mobility

Sprint

Binge On, Binge Off

Where are you likely to really be able to "Binge On" using T-Mobile's network?

T-Mobile's Binge On campaign offers customers the ability to stream certain video providers without incurring data charges. We wanted to see where customers might be likely to receive HD quality video with Binge On. Preliminary results suggest that HD quality on T-Mobile is much more likely in California's cities than anywhere else. Stay tuned for more in this area as we finalize our analysis.

More on How We Did This
Using the spring 2015 field test data, we've started to dig into video streaming quality for the four major providers. We created four categories, High Definition (HD), Standard Definition (SD), Low Definition (LD), and No Service (NS). The image above was created by designating each test location with a video quality category and then running a geo-spatial interpolation. The data were validated using a technique called "Cohen’s Kappa," which compares measured values with values predicted by the model.

The areas without color are either areas outside the provider's coverage, or are un-populated.

Definitions:

• HD video (720p)  >= 2.5 Mbps downstream with >90% of measurements being > 2.5 Mbps
• SD video (480p) >= .7 Mbps < 2.5 Mbps downstream with > 90% of measurements within that range
• LD video (320p) < .7 Mbps
• NS (no service) - no connection

Broadband Adoption Correlated With Income. Surprised?

Recently, several articles have featured studies on mobile broadband adoption and why more people aren't subscribing to wireline broadband. I thought it would be interesting look a little at these studies and revisit the CPUC's own study on wireline broadband adoption.

Telephone Landlines Are Disappearing

• In December, the Centers for Disease Control found that as many as 47% of American adults lived in homes with only mobile phones for telephone service (no traditional telephone line).
• More than 2/3 of adults aged 25–29 (71.3%) and aged 30-34 (67.8%) lived in households with only wireless telephones.
• The percentage of adults living with only wireless telephones decreased as age increased beyond 35 years: 56.6% for those 35–44; 40.8% for those 45–64; and 19.3% for those 65 and over.

Broadband Not Relevant To Low Income Americans?

• The Benton Foundation’s blog has a post by three researchers that suggests one of the key barriers often mentioned in broadband adoption, “relevance,” may mask other fundamental issues such as price and ability to pay. The researchers suggest in response to broadband adoption surveys that show people “not interested in getting online,” adding follow-up questions that focus on cost and digital literacy.

Income and Broadband Adoption

• The CPUC's California Broadband Report based on June 2011 wireline subscription data looked at seven variables to see if there might be any correlation with wireline adoption rates. The table below summarizes relationships between the wireline adoption rates of California census tracts and a series of demographic variables.
•  This list is not intended to be exhaustive, but rather includes variables with some of the highest explanatory power, such as median household income and educational attainment indicators, along with variables that would be expected to hold a high degree of explanatory power, but in fact do not, such as density.

 It is necessary to note that these variables are highly correlated not only with adoption rates but also to one another. This limits the ability to accurately determine the contribution of each individual variable to overall changes in adoption rates within a statistical model that includes more than one explanatory variable. Multivariate models can improve overall explanatory power, but interpreting the results becomes increasingly complex.

CalSPEED results used in latest FCC Wireless Competition Report

On December 23, 2015, the FCC released their Eighteenth Wireless Competition Report (DA 15-1487), and in it they use results from our CalSPEED tests and compare them with Ookla, RootMetrics, and the FCC's own speed test app results.

Based on the CalSPEED data, latency continues to improve. Verizon Wireless and Sprint have seen dramatic improvements in their latency over the last several reporting periods, and are closing in on AT&T for the lowest latency measurements. While T-Mobile has the highest mean latency, they have also demonstrated dramatic improvements during the last several test periods.