Friday, March 18, 2016

Monday Evening Sushi Buffet - T-Mobile's "Binge On" Reconsidered



As more people have dug into T-Mobile's Binge On service, it has become clear that it's not as great as it sounds. It starts to look like the Monday evening sushi buffet -- the fish market is closed on Sundays, and all that leftover fish from Saturday night must be sold! Which would you prefer, a $50 steak dinner, or a $10 all-you-can-eat sushi buffet? Depends on your budget and what you're willing to put up with.

My earlier analysis of where you might be likely to stream high definition video using T-Mobile's Binge On service did not take into account the fact that streaming HD video with Binge On may be impossible, because T-Mobile throttles the speed down to 1.5 Mbps. The Electronic Frontier Foundation did a test using Binge On and determined that their video optimization is "just throttling," and that T-Mobile applies the speed throttling "indiscriminately to all video."

Does that make the streaming video coverage map in my earlier post wrong? No, but the only way you will likely be able to receive HD quality streaming video is by disabling the Binge On feature.

Another way of looking at Binge On, suggested to me by a colleague: T-Mobile is simply "incentivizing lower quality video usage … lower load on the network [coupled] with kinda an acceptable level of quality."


Wednesday, March 16, 2016

Coachella Live: Will My Phone Work?

For two weekends next month (April), a bunch of bands will be playing Coachella at the Empire Polo Club in Indio, California. In case you or someone you know is going, check out the mobile broadband maps to see what level of service you might expect in that area.

T-Mobile, which tends to have weaker rural coverage than either AT&T Mobility or Verizon Wireless, nevertheless appears to have strong coverage around the Empire Polo grounds on 51st and Monroe. Maybe that's not surprising given that T-Mobile's target demographic is the youngest among the four mobile providers. For Sprint, our model suggests there is no service in the area of the concert venue, but we encourage concert-goers who have Sprint to download CalSPEED (free download from iTunes or Google Play) and run some tests for us, because those test results will help us improve our model. The one CalSPEED test we have from crowdsourcing for Sprint in downtown Coachella shows no service, but a field test location near Harrison and Airport Boulevard shows over 7 Mbps down and over 4 Mbps up.


Here are images showing the interpolated mean minus two standard deviation downstream speeds for each provider for the area around Coachella. Concert venue is indicated by a star.

 T-Mobile: 3.0-6.0 Mbps downstream


Verizon Wireless: 3.0-6.0 Mbps downstream

AT&T Mobility: 1.5-3.0 Mbps downstream

Sprint: No service

Monday, March 14, 2016

iPhone vs. Android - Which Is Faster?

Earlier this year, we devoted a week of testing to compare the iPhone 6 with the Samsung Galaxy S6, and the Samsung Galaxy Tab 4. You might expect the iPhone to be the fastest for all providers, but it was true only for T-Mobile downstream. Below are graphs comparing the mean downstream (sending) and upstream (receiving) speeds for each (in blue), as well as the average standard deviation (in red).


We found that the GPS readings for the iPhone were slightly off. We know this, because all three devices for the four providers were tested in a stationary location each time. The plot above shows how some of the iPhone locations appear slightly off from the others.





Friday, March 4, 2016

Bicoastal Speeds Are Converging: Sneak Peek At Upcoming Novarum Report



I've touted the benefits of CalSPEED's use of servers on both coasts of the U.S., and so you may be asking, "How's that working out for you?" It's showing an interesting trend: convergence. In the upcoming Novarum report on the Spring 2015 mobile field test results, Ken Biba shows how speeds to west and east coasts are converging. In the graph above, we see the difference in speeds for any location where user tests to both the west coast server and east coast server. For Sprint, the east coast speed is 10% slower, and for Verizon, it is about 20% slower.

From the Novarum report:
"In the best possible time, the physics of data transmissions adds about 80 milliseconds of additional latency to get from one coast to the other, in addition to any local wireless access latency. In the case where the  latency difference between servers on opposite coasts is zero, we speculate that traffic for both servers is peered through a geographically central location, such as Kansas. The latency difference between East and West servers has decreased over time - converging on 80 milliseconds."