Category Archives: jim warner

CPUC’s second field test building consistent picture of mobile broadband performance

by Steve Blum • , , , ,


Spring up, Fall back.

More mobile broadband performance measurements are available and accessible to Californians, thanks to field testing done by the California Public Utilities Commission and mapping and analysis done by Jim Warner at U.C. Santa Cruz.

Warner, who is a network engineer for the University and chair of the Central Coast Broadband Consortium’s technical expert group, took the data collected in the CPUC’s first and second rounds of mobile data field testing and fed it into Google maps. The CPUC took readings at 1,200 locations in each round, for the most part repeating measurements at the same locations. AT&T, Sprint, T-Mobile and Verizon service was tested using both a smart phone and a netbook.

“It is comforting to see that in some speed tests, Verizon results were above 40 Mbit/sec – fully six times faster than the California broadband definition,” Warner wrote. “This means that the equipment used for the tests is not coloring the results.”

Warner shows, as an example, Verizon has reasonably consistent improvement in the Monterey Bay Area. On the other hand, to my eye there’s little difference in Verizon’s coverage in the Gold Country.

Along I-80 in Solano County, to take another example, some of AT&T’s results are better, some worse. That supports Warner’s observation that “tests were done throughout the day at times that were not controlled for network loading by other traffic. Tests done early in the day might face less cross traffic congestion than tests later in the day.”

In other places I checked, the variance also went both ways. That tells me that something other than new construction is the cause. But you can look for yourself and come to your own conclusions.

More unlicensed spectrum on the way

by Steve Blum • , , ,

Super-sized WiFi too.

The Federal Communications Commission (FCC) took a big step towards fulfilling a promise that Chairman Julius Genachowski made at CES in Las Vegas last month. Yesterday, commissioners voted unanimously to start the process of opening up 195 MHz of spectrum in the 5 GHz range to unlicensed uses such as WiFi. (H/T to UCSC’s Jim Warner for tipping me off).

“WiFi congestion is a very real and growing problem,” Genachowski said. “Like licensed spectrum, demand for unlicensed spectrum threatens to outpace supply. The core challenge is the dramatically increased use of wireless devices, which require spectrum.”

The proposal under consideration would open up the 5.35 to 5.47 GHz and 5.85 to 5.925 GHz bands to unlicensed users. It’s not just about more bandwidth. The move would also bridge current allocations and create a contiguous unlicensed band from 5.15 to 5.925 GHz, opening the door to wider WiFi channels and faster speeds.

“What excites me about today’s Notice of Proposed Rulemaking is that we are building on these past successes and using spectrum ideally suited for unlicensed use,” said Commissioner Ajit Pai. “Enhancing the contiguity and size of the 5 GHz blocks contemplated in the item should allow wider channels for higher bandwidth transmissions. For example, a 160 MHz-wide channel could deliver 1 gigabit of data per second. That’s Super Wi-Fi.”

If eventually approved, the new rules would have to allow for frequency sharing with some licensed users, so different restrictions on metrics like power and antenna characteristics would likely be placed on different part of the new band. But that’s for manufacturers and standards groups, like IEEE, to work out.

Full text of the notice is here.

FCC crowdsourcing mobile broadband measurements

by Steve Blum • , , , ,

Can y’all hear me now?

Via Jim Warner, U.C. Santa Cruz network engineer and chair of the Central Coast Broadband Consortium’s technical expert group: the Federal Communications Commission (FCC) will ask the public to download an app and take mobile broadband coverage measurements…

The FCC and its contractor SamKnows will soon be announcing a program to collect crowd source data on mobile broadband performance. A program will trigger tests like iperf and ping and report the results along with handset and location information to a central database. Parameters from the phone that will be reported include bearer (HSPA, CDMA, LTE, WiFi, etc), reported signal strength, tower ID and measurement time stamp.

The opt-in testing app will come in iOS and Android versions. The Android app will be beefier, presumably because Apple locks down access to geek-level system resources. The FCC started working on this project in September.

The California Public Utilities Commission (CPUC) is taking similar measurements, to validate (or not) the coverage claims made by mobile carriers. While CPUC is doing the work in-house, the FCC wants you and a million of your closest friends to submit test results…

This differs from the California survey in that all data is contributed by users so there will be vastly more of it. A plus for the California approach is that it carries no risk of exposing subscriber’s private information since all testing is done by employees. The FCC has not yet settled on a plan to fuzz location information to protect user privacy.

Consumers might or might not have to pay for the bandwidth they use to do the testing…

Users who opt in to the measurement program may be contributing some of their monthly plan quota to the testing. The FCC has asked carriers if they can exempt traffic to measurement nodes by their IP addresses so that testing carries no user-side incremental cost. The risk here is that announcing the test server addresses to the carriers would give them the information they would need to prioritize test traffic and improve their scores. Of course, the carriers will know the addresses of the test servers anyway, since this information must be public.

The full text of Jim’s analysis is here, and he’s also posted the presentation that the contractor, SamKnows, made to the FCC.

The program is still under development, with a working group meeting every three weeks or so. I’m looking forward to seeing the results – should be very interesting. Not as much fun as crowdsourcing radar traps, though.