Imagine you're watching PBS right now and Bill Nye The Science Guy comes on. Suddenly Bill Nye's disembodied head is floating across your screen, with 1s and 0s streaming in and out of his ears.
"Did you know," shouts Bill Nye (I don't know why he's shouting, but he is), "that when you make a cellular phone call your voice is transmitted as digital data, just like how a CD or DVD stores information?"
You change the channel. You knew this already, Bill. Everybody knows this. And yet when you make a phone call from your couch, to tell your friend that Bill Nye is "losing his edge," the digital data of that phone call isn't routed over Wi-Fi through your home's fat internet pipes. Instead your carrier confronts some laws of physics and gets those 1s and 0s from the little brick in your hand to a cellular tower that's miles away, which then sends the call over a mess of wires, pipes, and lines that dates back to Alexander Graham Bell himself. It's all very impressive, but who are they trying to impress? Certainly not your dormant Wi-Fi router, certainly not your bitrate-weathered ears, certainly not Bill Nye.
Unless, of course, you use T-Mobile's Wi-Fi calling service. Fun fact: T-Mobile has a service called "Wi-Fi calling." Have you ever heard of it? T-Mobile hopes you haven't.
Wi-Fi calling was originally launched as "HotSpot @ Home" in 2007 by T-Mobile as a bid to compete with the landline. It was a reasonable idea, after all: by providing a reliable service inside the home, a cellphone could replace the stalwart landline, whose only technological advantage in 2007 was reliability and voice quality. Built on the little-used UMA standard, the service shipped on a couple of specially Wi-Fi-equipped mid-range featurephones and was promptly forgotten.
UMA stands for "Unlicensed Mobile Access," and describes any method of taking a voice or data connection and sending it over an "unlicensed" internet connection (anything that's not the carrier's own network) straight into the carrier's nerve center, where it's switched into the regular phone network. Security is maintained due to the unique SIM in each GSM phone.
After HotSpot @ Home launched in 2007 to little fanfare, as a $10 monthly landline-killer, two things happened to vindicate T-Mobile's strategy:
1. Apple launched the iPhone, which marked the beginning of the "smartphones are for everybody" era.2. Everybody decided they could replace their landline with a cellphone after all.
Unfortunately, these two trends, while growing in parallel, haven't complemented each other very well in the past five years.
The physics problem
Have you heard of Wi-Fi calling? T-Mobile sure hopes not
T-Mobile remains the lone major proponent of Wi-Fi calling in the US, despite the fact that UMA is an open standard (which AT&T, as the other GSM carrier in the US, could easily implement), and the fact that there was a CDMA version of UMA in the works as far back as 2005. But even T-Mobile is a little bashful in promoting its service, confirming to me that the feature would "never" be a top-line of an ad campaign.
Of course, having Wi-Fi calling, advertised or not, is a big leg up that T-Mobile has over other carriers. When you set up a Wi-Fi calling-enabled device, the phone will inform you of the functionality and ask you if you'd like to use it. Many people I've spoken to were unaware that T-Mobile offered the service until they actually started using their device — and now find they can't live without it. It's refreshing that the carrier is taking the issue of in-home calling seriously, but there seems to be a limit to how far it's willing to go, and how indispensable it's willing to make Wi-Fi.
It turns out, the first step is admitting you have a problem.
"To implement UMA may take a certain level of humility," says Sascha Segan, Lead Analyst at PC Mag. "It means admitting that your network isn't perfect, admitting that there are places, a lot of places, where people have stronger Wi-Fi signal than they'll have cellular signal."
See, the cellphone networks didn't ever beat out landlines for in-home reliability — a huge concern at the beginning of this century — it's just that people decided a cellphone was "good enough" at some point, and so maybe they didn't bother to install a landline when they moved to a new house, or didn't renew that nasty long distance plan when they were making up their family budget for 2008. It doesn't hurt that a typical cellphone plan today includes $30+ of data charges that weren't common in 2007. Something had to give.
I spoke to Josh Lonn, T-Mobile's director of product marketing, at length about Wi-Fi calling, and while he was very forthcoming and bullish about the service, he went to almost comical lengths to make sure he didn't slip up and in any way imply that the existence of Wi-Fi calling means there are holes in T-Mobile's "all around great 4G experience." While I kept on trying to slip the word "augment" into my questions, Josh and his team kept firing back with the word "complement."
But even Josh, like his best-value nationwide 4G network, makes mistakes sometimes. "The reality of radio networks in general is that there's no such thing as a perfect radio network," Josh confessed, in his moment of weakness. "It's impossible and that's driven by physics, how cell sites work, economics, land use restrictions, and practical aspects like 'you just can't put a tower in everyone's backyard.'"
Exactly.
Of course, every carrier has its own coping mechanisms for service problems, typically activated at the point the customer complains. If you say "I don't get good reception in my home," for instance, T-Mobile will inform you of your phone's Wi-Fi calling functionality (if your Android device has received the appropriate update), while the other carriers will likely offer you a femtocell device. Now even T-Mobile is offering a femtocell fallback, which serves as another ringing non-endorsement of UMA.
Femtocell is sort of like the Rube Goldberg take on Wi-Fi calling. Instead of routing your voice call directly through the internet, a femtocell device pretends to be a real cell tower, which captures your phone calls and then... routes them over your home internet. Cost is a bit of a crapshoot: if you don't wrangle a freebie from a kind customer service rep, some of the carriers will actually charge you for the device — which can be as much as $350.
"There's no such thing as a perfect radio network."
As odd as these Wi-Fi-avoidance contortions might sound right now in the early days of 2012, they're going to get really silly as Verizon and AT&T implement "VoLTE" (Voice over LTE, pronounced "voltie") later this year — in fact, Verizon allegedly has a couple trial VoLTE markets up and running.
With VoLTE there's no difference between "internet data" and "voice" anymore; it's all IP. While the difficulty of UMA implementation, or the lack of a finalized CDMA alternative to UMA, might've worked as an excuse before, VoLTE means the excuses are running out.
After a call is made digitally between your phone and the cell tower, the next step is a bit of a mess. There are good odds it will be carried over some sort of fiber backbone, but there are plenty of legacy elements sitting around in the phone system — some decades old — that could become involved. It works, sure, but it's anachronistic. A ship will still get you across the ocean, but why not just fly?
While connecting a call over UMA can have its own problems (like latency, or Wi-Fi interference), there aren't any jumps or transfers the call has to make — it's just a straight data transfer between the phone and T-Mobile's data center. That means that nothing in that legacy tangle of phone system can get in the way of functionality or quality of service that T-Mobile wants to provide, and, of course, it means that the best pipe wins.
Once calls on Verizon and AT&T are made over VoLTE, there's little stopping them from being re-routed over Wi-Fi as well, at the mere flip of a switch. The day a carrier offers me an LTE femtocell is the day I weep for a cellular radiation-riddled humanity.
Fighting the power
Meanwhile, on the other side of town, is Republic Wireless. Republic is a new MVNO (mobile virtual network operator) from the folks at Bandwidth.com, which rents its cellular network from Sprint, but whose main goal is to teach the world to Wi-Fi call.
"We're the first people in the market to come to market with a plan that uses Wi-Fi first, and cellular as the backup," says Brian Dally, the general manager of Republic. "Every other mobile carrier talks about offloading to Wi-Fi, we talk about failing over to cellular."
While offering a nearly identical product to T-Mobile, Brian is the anti-Josh Lonn. "We theorize in our model that people should be able to offload perhaps 60% of their cellular footprint," he claims, which makes Republic's dirt-cheap $19 monthly unlimited plan possible. ("It's not $19.99 by the way, it's $19," Brian is sure to point out).
There's a real fight-the-power strain that runs through Brian, in fact, which matches up well with his MVNO's name: "It's not that voice calls or text messages are particularly expensive to move on the 3G network... What's happening is that carriers are exerting pricing leverage for moving those calls and text messages. And the consumer is on to them."
He's not finished, either: "The cellular operator has no clothes because now Wi-Fi is a free and open resource for many, many people."
Unfortunately, Republic Wireless isn't ready to confront any emperors yet. Its service is still in a literal beta, and its current incarnation has the telltale signs of an ideological movement that hasn't yet become purely practical. Sure, you're only paying $19 a month, and you're really sticking it to the carriers, but you're also using an LG Optimus. The phone was low-end when it was released over a year ago, but it's laughably outdated now, with a small, low-res screen and a completely generic hardware shell. Brian promised me some better devices by the time the service is ready for public consumption.
The service has plenty of rough edges as well, but at least the core idea is present. While voice quality is a little shrill (not helped by the phone's speaker), I could make calls from anywhere in my apartment for hours on end without the regular blips in quality or dropped calls that are typical for me on all four carriers.
It's sort of like being the first of your friends to get a Nomad Jukebox and painstakingly ripping your MP3s and sorting your whole music collection into virtual folders — you might be participating in the future of music consumption, but you're also unlikely to be having very much fun, and you'll look like a total square.
"The cellular operator has no clothes..."
"With Wi-Fi we all build the network, and we all get to make calls."
Still, that future is tantalizing. After years of conditioning, it's actually kind of an odd sensation to make a call without problems — like watching a movie in a theater after looking at stuttery YouTube videos on a bad internet connection all day — and it's almost enough to make me overlook the phone's many flaws as a smartphone. Still, I take my iPhone with me when I leave the house: I wouldn't be seen dead with the Optimus up to my ear.
Even T-Mobile, whose service is much more mature with some truly stellar voice quality, has trouble getting Wi-Fi calling onto its hottest new devices. The latest tactic is to launch phones as soon as possible, and then add Wi-Fi calling after the fact, in the form of an OTA update. Our review unit of the HTC Amaze, for instance, just got the OTA update, a few months after launch. The take-your-time approach is a luxury of Wi-Fi calling being a non-advertised feature, and to T-Mobile's credit, every smartphone gets Wi-Fi calling eventually, but for anyone who has come to rely on Wi-Fi calling it's surely frustrating.
A company called Kineto Wireless builds the software for T-Mobile, and then hands the handset manufacturer the necessary code to put in the Android kernel, which the Wi-Fi calling software hooks into to make secure calls. "This isn't an app that can be downloaded from the marketplace," explains Steve Shaw, VP of marketing at Kineto. "It comes pre-loaded on the phone and that has to do with primarily security." Much of the difficulty comes from doing SIM authentication over Wi-Fi, something most apps on the phone aren't allowed to touch.
While the software change is a minimal one, it's another hassle on top of the endless carrier customizations that manufacturers already do. "The last thing they want to do is embed a little bit of code specifically for one feature for T-Mobile and it throws off the entire schedule for the global or regional launch," says Steve.
Meanwhile, Republic Wireless rolls its own UMA-like software compatible with Sprint's CDMA network, with Bandwidth.com's VoIP mojo underneath. While T-Mobile has to deal with the politics of high-profile multi-carrier, multi-market device launches, Republic has to deal with Sprint. There's no alternative to the Big Four if you want nationwide cellular access, no matter if you call it a "fallback" or not, and even MVNOs who rent access must have their devices of choice provisioned by big papa carrier.
"When it comes to handsets we're like every other consumer: we have to follow the rules that Sprint-Nextel sets," says Brian. "They built the network, they make the rules. That's why we're emphasizing this alternative network based on Wi-Fi, because with Wi-Fi we all build the network, and we all get to make calls."
In case it wasn't obvious, Brian and his friends at Bandwidth.com aren't just about changing how you make calls, or, to use his hyperbolic phrasing, making sure "we all get to make calls" — they'd like to shake up the whole industry.
Reality check
The carriers certainly do make plenty of rules, but while only a select few phones from T-Mobile and Republic can make authorized, bona fide-phone-number calls over a regular Wi-Fi connection, almost every smartphone on the market has another way.
It's called "Skype." Or maybe it's called "FaceTime." I've heard people refer to it as "Google Voice" on the odd occasion.
See, whether carriers like it or not, people make IP calls all the time, from their computers and from their phones. Even from their iPods. In fact, the only reason T-Mobile, Republic, and Kineto play these crazy Android kernel contortion games is because we're all so fond of making calls to and from our phone numbers.
Here's the equation for making the switch, as I see it: as the carrier's charges approach infinity, the likelihood I'll do the necessary legwork to find a calling alternative nears one. Why pay AT&T $7 a minute to call my friend in Germany when Skype will do it for free — with video, to boot!
A typical voice call tops out at around 12kbps
Kineto's Steve Shaw thinks AT&T's recent launch of its AT&T Call International App, which lets you use your smartphone to make VoIP calls to real phone numbers in various countries at just-barely-more-expensive-than-SkypeOut prices, is actually a reaction to Skype. "Mobile operators haven't traditionally seen a Skype app running on your phone as a threat," says Steve, "but they're starting to, so they're looking at new things they can do."
With or without reactions from carriers, it's clear that Skype has fundamentally changed how we communicate. There was a time in my life, not too long ago, when I had never made a call to or from an international phone number. If I had, I probably would've bought a calling card instead of subjecting myself to horrific carrier rates. Nowadays, calling another country is trivial with Skype, either client-to-client, or with Skype's low per-country rates for calls to regular phones. I've played hours of StarCraft with someone in Australia while chatting over Skype, and didn't even bat an eye — I certainly didn't have to ask any of the Big Four for permission.
Given the seeming inevitability of more-or-less-free internet calling on their devices, the carriers are playing a game of chicken, seeing how long they can hold onto their current revenue models — which emphasize voice minutes and SMS messages, rather than data. I understand the impulse: a majority of their revenue (somewhere between 50 and 80 percent, based on who you ask) still comes from voice and SMS. Still, the model is becoming increasingly separate from reality.
A typical voice call tops out at around 12kbps, but can scale as low as 4.75kbps when the going gets rough — silence is transmitted over GSM at a rate of 1800 bits per second. For comparison's sake, the telegraph transmitted information at a rate of 150 bits per second, back in the 1850s. Meanwhile, the LTE specification lists speeds of 300Mbps down and 75Mbps up. While those numbers are theoretical maximums, which will slowly be built towards as LTE matures in the real world, Verizon Wireless says that users should expect roughly 5Mbps from its service right now — which we far exceeded in our own tests.
I'm not good at math, but it seems to me like you could carry on a few hundred voice calls simultaneously over an LTE connection. "Oh, but what about HD voice calls, the future of phone calling?" you say. Bump the bitrate to... 12kbps. That's what the AMR-WB (Adaptive Multi-Rate Wideband) codec calls for. Maybe 30 or 40kbps if you want to add in some background noise cancellation and stereo audio. Older, less efficient codecs (often in use by VoIP services) hover around 60kbps. Even Skype's vaunted voice codec tops off at 40kbps, going as low as 6kbps for low-end devices. There's so much room left over for 1080p YouTube streaming, trust me.
Our devices and apps have innovated wildly inside of this new 3G and 4G headroom, while voice quality has stagnated for decades. I have trouble understanding people on the phone — even on the rare occasion when I have a solid signal. It was bad enough when I had a landline, but cellphones have really made voice calls an exercise in frustration. I typically prefer SMS when I actually need to communicate with someone.
"Carriers have been having a race to the bottom with voice codecs and call quality for years."
For me it's a minor hangup, but it's a serious issue for the millions of people with hearing problems, or just slightly limited hearing (like most old people). It's not even a question of "good enough." The current audio codecs don't even support the basic range of human voices, much less the wide range of human hearing. And this isn't a new problem.
Example 1: Typical narrowband sample
Example 2: AMR-WB sample (source)
"There are a lot of people out there who feel strongly about the quality of their voice calls," says Sascha Segan, "it's just that the carriers have been having a race to the bottom with voice codecs and call quality for years."
A standardized wideband "HD voice" audio codec, with vast improvements over current voice quality, was approved in 1987. That's 25 years ago! Even AMR-WB, the hot codec of the moment, is a decade old standard at this point. The first live tests of AMR-WB on a cellular network happened in 2006, and yet we won't be seeing HD voice in the US until Verizon rolls out VoLTE later this year.
A standardized wideband audio codec was approved in 1987
The bandwidth for these calls has been there for years, and when a call is made over Wi-Fi it's trivial. Orange in Europe is one of the first carriers to offer an HD voice service over 3G cellular (its first markets launched in 2009), and Orange also just-so-happens to have a robust Wi-Fi calling service, powered by Kineto. From Kineto's point of view, there was no work to be done: once the call is routed over Wi-Fi, it's all data, and higher bitrates are still just data.
Republic Wireless says that Bandwidth.com already has HD voice VoIP service, "but the cheap Android hardware we're using today doesn't support that codec." While he didn't come out and say it, Brian seemed to imply that the higher end handsets Republic hopes to offer soon should be able to support HD voice.
And, of course, once a "bleeding edge" user like me finally manages to get a phone service in the US that will serve HD voice calls, how much longer will it be until those calls will connect through to an HD voice user on the other end on a regular basis? Another 25 years?
Dumbing down the pipe
I'm sick of fighting physics every time I make a call from the couch
If it's all, at the end of the day, about getting me to overpay for a phone number, then why not just get rid of the phone number — or at least marginalize it? The irony is that the LTE networks these carriers are working so hard to build just might give us, the people, a chance.
If a service like Skype or FaceTime — both of which can work great over an LTE pipe — could offer near-perfect ubiquity and interoperability, in addition to the implicit cost-savings, quality leaps, video functionality, and Wi-Fi resiliency, then legacy voice calling would have a real fight on its hands.
Kineto's Steve Shaw sees the window of opportunity there: "I think the mobile industry is all excited about running to LTE, but in the back of their minds they're thinking 'oh my god we're gonna give people this really high speed, really low-latency IP connection directly to their mobile phone and are we really gonna be able to meet the needs and give consumers everything that they want over the LTE network?'"
For its part, Kineto is working to help carriers "bridge the gap between LTE and Wi-Fi." While UMA won't be relevant for long, the need to get high-bandwidth activity off a burdened network and onto Wi-Fi is getting hotter than ever. Physics still apply.
There's another company called Ruckus Wireless that's working with carriers to offload data traffic to Wi-Fi, creating networks that the carriers can still monitor and control. Meanwhile, a company called Devicescape has a crowdsourced list of open hotspots, and is about to partner with an unnamed tier-1 carrier to deploy its service to a multitude of data-hungry handsets. Both companies are looking to change Wi-Fi from something users have to worry about and manage, to something the carrier figures out for them.
"We are facing a spectrum crunch," says Sascha, "the carriers are probably going to face capacity problems; they should get serious about Wi-Fi."
Still, Sascha isn't ready to throw away his phone number and flip off the carriers just yet: "Hackers, Geeks, people on serious budgets, are going to cobble together complicated experiences, but something is going to have to be really really compelling to draw people away from picking up the phone and dialing a number wherever you are."
T-Mobile feels safe in the face of a Skype onslaught as well. "Anytime you have a one click or two click to get to a feature, no matter how compelling that feature is, it's going to be a barrier to the majority," says Josh.
They both have a point. While in theory it's not more work to call somebody over Skype instead of the legacy phone system, it's not an integrated experience on many devices yet. Meanwhile, FaceTime embeds itself nicely into regular contact cards, but as long as it's iOS-only, it can only ever be a "complement" to regular phone service, not an "augment."
Still, Apple and Microsoft (who now owns Skype, though it still hasn't shipped an app on Windows Phone yet) probably have the best chance to pose a serious threat to carriers in the realm of calling, should they wish to. Apple is already working to get the iPod touch, running FaceTime and iMessage, to be seen as a phone alternative for tweens, and I imagine Apple could get a ton of traction pushing both FaceTime and iMessage (and a voice service, of course... VoiceTime?) as open standards. Supposedly Apple even considered building its own phone network on Wi-Fi, before it launched the iPhone. Here's a chance to make Steve's dreams come true.
The best things about phones aren't provided by the carrier
Meanwhile, Microsoft has the exact opposite to-do list: it has an app that's already ubiquitous and cross platform, but it needs the guts to push Skype as a phone call alternative, with Windows Phone as the premiere platform, which would mean sticking it to the carriers in a big way.
"If Microsoft integrates Skype more tightly into Windows Phone 8, you could see something really disruptive happening," says Sascha.
But Microsoft needs all the help it can get from its friends at the Big Four right now, so don't expect an aggressive move anytime soon.
Even without a big play by Microsoft, Apple, or Google, however, services that don't rely on a phone number are only growing in popularity. In places where price or functionality are a concern, you see IP services popping up everywhere. Businesses are replacing huge, spendy phone systems with desktop Skype clients, grandparents are FaceTiming their families instead of calling them, and video chat is a boon for people with hearing disabilities. Vonage just rolled out Vonage Mobile, in a big move to compete with Skype, and Dell just launched a phone-number-attached VoIP service in Canada called Dell Voice. T-Mobile is doing something like Google Voice with its Bobsled service. Change is coming.
It only takes a cursory glance at your phone's homescreen to see that most of the best things about phones aren't provided by the carrier. This isn't a new idea: the "everything except make calls" joke about the iPhone has provided the most material to late night stand-up routines since Clinton-Lewinsky.
Of course, all of those glorious applications are powered by the impressive, multi-billion dollar pipes that carriers build for us. Nobody is arguing that carriers shouldn't be compensated for that service — in fact, I think the current amount I pay AT&T for voice, data, and tethering sounds about right. If somebody wants to offer that stuff to me cheaper, then that's great, but I'm not complaining about what I'm spending. I just wish carriers would realize where their actual value lies, and stop standing in the way of progress.
The big bad bugaboo feared by carriers is becoming a "dumb pipe." The fate of the ISP. Alright then, be a "smart pipe."
You know what a smart pipe would do, as its very first super special job? It would offload phone calls to Wi-Fi. I'm sick of fighting physics every time I make a call from the couch.