1. Global Internet Interconnection Conference Walter Johnston Chief, EMCD Office of Engineering &Technology FCC The Views Express Herein Are Those of the Author and Do not Necessarily Reflect Those of the FCC

2.  Broadband and the Internet empowering our society ◦ Productivity Gains ◦ Enhanced communications ◦ Entertainment ◦ Public Safety ◦ Connected Government  Important to measure availability and quality of broadband ◦ Availability ◦ Performance ◦ Year by Year Trends  Open Internet Order established transparency requirements for Broadband Internet Access Providers (BIAS)

3.  2010 – FCC issued National Broadband Plan ◦ Established goal to obtain Internet performance data to support policy development  2010 – initiated program to provide most accurate data on fixed broadband to the home ◦ Cooperation of major ISPs supporting 85% of consumers  13 – 15 ISPs in program; remaining 3000 ISPs support 15% of population  2014 – initiated coverage of mobile broadband performance ◦ Crowd sourcing approach ◦ Data collected includes performance data, cell tower/service coverage data, and phone data

4.  Instrumented ~6000 consumer homes in U.S.  Annual report ◦ Continually collect data from these homes  Speeds, latency, loss ◦ Baseline data for report collected Sept/Oct annually ◦ All collected data available on website  Research initiatives supported when feasible ◦ WTF – Georgia Tech ◦ Latency data provided to CAIDA ◦ CGN studies, congestion studies under consideration ◦ Consumer nodes can be limiting factor  Limits/Opportunities established by WTF  Recently finding 100 Mb/s rates and higher rates challenging measurement platforms ◦ Hope to test 1Gb/s services this year (Beta)  Interest in shaping program to better support research

5.  Over first 3 years, ISP services improved ◦ Ration of delivered/advertised increased ◦ Spread between best/worst diminished  Provided better understanding within Commission regarding functioning/operating of the Internet ◦ Supplements additional information sources  Provides enduring baseline data for Commission ◦ Information integrated into other FCC programs

6.  Started analyzing service consistency ◦ Streaming services changing landscape ◦ QOS Performance over time  Video streaming performance ◦ Service characterization: Netflix, Hulu, YouTube ◦ Maximum achievable streaming capability  Interconnection points ◦ Supporting research using latency to characterize congestion ◦ Hope to expand data program focused on Interconnect  FCC Technological Advisory Council ◦ Recommended in 2015 expansion of program ◦ What network points to capture ◦ Where to evolve ◦ How to obtain data

7.  Crowd sourced data collected from volunteers’ phones ◦ Open methodology developed with industry collaboration  Clients available on GitHub as open source for inspection ◦ No PII data collected, no ads, no commercial intent  We have spent two years verifying validity of data and examining how best to present

8.  Android ◦ Released in mid-November 2013 ◦ Available at Google Play: FCC Speed Test  iPhone ◦ Released February 2014 ◦ Available to public at iPhone App Store: FCC Speed Test  For a very short time we were 4 th most popular app in the store  Android phones provide 4x data as iPhones due to automatic scheduling of tests  Favorite comment ◦ “Very useful app and surprisingly not ugly”

9.  Event oriented;  News article or blog post last about 10 days  Evolving outreach program to keep reinforcing program and apps  We have been silent on this app for two years due to industry concerns ◦ However – last two competitive wireless reports included MBA data ◦ Will seek to spotlight program in 2016

10.  Extensive analysis of data with comparison to other data sets ◦ e.g. Ookla, RootMetrics  At sufficient levels of aggregation ◦ Data meets statistical validity criteria ◦ All data sets in general agreement  At sufficient levels of disaggregation, weaknesses in all methodologies become apparent ◦ Crowd sourcing and drive testing both dependent upon limited sample sets at high area resolutions

11.  Huge datasets when aggregated  Sparse datasets when data disaggregated ◦ Drive testing: specific locations measured twice a year ◦ Crowd sourcing: a few consumers as proxy for all  Street level measurements mean what to average consumer?  No ‘average consumer’ at high resolution  How to present data challenging problem

12.  Mobile competitive environment is fierce  Government comment on performance perceived as threatening ◦ Many service providers advertise ‘best network’  Vigorous push-back from some participants  Lengthy process to establish validity of data ◦ But also a recognition that fairly presenting mobile information to the public is difficult ◦ Mapping turned out to be a problematic approach

13.  Privacy aspects of Internet data a growing concern ◦ NTIA/FTC have led industry discussions  Privacy discussions still in early stage  Many Company's T&Cs seek to protect the company, not the consumer  FCC collection of data  Privacy policies developed with outside experts, wireless carriers, FTC and FCC ◦ Goal to protect consumer ◦ Seek only necessary data supporting understanding of broadband services and broadband network  Committed to open data release subject to 3 rd party review for privacy compliance

14.  Handset data fully anonymized ◦ Led to changes in initial design strategies ◦ Privacy analysis of data with outside parties ◦ Further obfuscation of data when required for public release  Thresholds, aggregation of some parameters  Expect to begin releasing data in 2016  Other agencies and organizations collecting mobile data ◦ Use of such data may be entangled with privacy concerns and sharing may be difficult  Goal to start a discussion on common privacy policies for collection of mobile data ◦ Non-aligned and sometimes conflicting interests in privacy objectives: consumer protection, industry market info, national security

15.  Began serendipitously -initially focused on producing a snapshot of broadband performance  Evolving towards ◦ Baseline studies ◦ Incorporation of addition critical measurements/network nodes  Increasingly aligning program to support Open Internet goals  Remain open for guidance on how best to evolve

16. Walter.johnston@fcc.gov