In the comments of the article about the analyst positing $9 billion for LTE deployment, many people were puzzled on why everyone is so focused on T-Mobile needing to deploy LTE. The obvious marketing answer is that everyone else has it, so T-Mobile needs to be competitive and deploy it to be able to say that they’ve got it too. By itself, that is an incredibly stupid reason to spend billions of dollars to deploy and maintain a brand new network.
However, there are very real and important reasons for deploying LTE. Before I get to that, a little history is needed to understand one very important thing: the Americas spectrum configuration was never optimal for the way we use mobile technology today.
The problem that caused this is rooted in how the first cellular networks were set up. The United States had the first cellular network in the world, and so the FCC had no precedent at the time to set up frequencies. They had to work with what they had and what they knew. And what they had was 850MHz, now known as the Cellular band (sometimes notated as CLR or Cell). What they knew was that mobile phone technology at the time only supported calls, and required large blocks to minimize interference. The Cellular band ran the first fully standardized analog cellular network, AMPS (developed by Bell Labs and Motorola). The licenses for the Cellular band were essentially handed out to two companies in each market, an “A” company and a “B” company (as the FCC called them, in reference to the 25MHz A block and 25MHz B block licenses). Over the years, the cellular market remained largely confined to the business elite because prices were really high back then. We’re talking thousands of dollars for a cellphone that could just barely make calls.
Over time, these companies partnered together and later merged together to form larger companies. With the exception of SunCom (which merged into T-Mobile USA) and several other smaller rural network operators, nearly all of those operators merged into the behemoths known as AT&T Mobility, LLC (which goes by AT&T) and Cellco Partnership, Inc. (which does business as Verizon Wireless).
A quick note: In 1990, all the European network operators finished developing the unified TDMA-based digital standard GSM to upgrade their disparate and incompatible analog networks to. It was published and they quickly migrated to GSM over the next few years. Around the same time, the European Union formally came into existence.
In 1996, Congress revised the 1934 Telecommunications Act to give the FCC broad powers on radio licensing and auctions to improve competition in the cellular phone industry (which had been full of corruption and collusion). The result was the first frequency band auction in the United States for the Personal Communication Service band, or PCS band for short. PCS had the mandate that only digital cellular networks could be run on the band. As a result, Motorola and Qualcomm each began developing digital technologies. Motorola’s efforts resulted in D-AMPS. That system is what Americans and Canadians colloquially referred to as TDMA. Qualcomm’s efforts, spurred on by the declassification of the CDMA air technology by the American military and with assistance from the military in developing the solution, led to the development of cdmaOne. This standard would later be finalized as CDMA2000. PCS was divided into several 10MHz and 20MHz spectrum blocks, as it was felt that the emerging digital cellular systems could use spectrum more efficiently and not all companies would be able to afford large 20MHz blocks of PCS. The incumbents at the time (A and B operators), as well as new entrants eagerly bid for new spectrum. The bulk of it went to telecommunications incumbents with Cellular band spectrum, but some new telecom companies got it too. This set the framework for future spectrum auctions. The companies that later became T-Mobile USA obviously participated in this auction.
Another quick note: In 2001, NTT DoCoMo took the CDMA2000 technology that Qualcomm developed, improved the technology, and refined it to become the W-CDMA technology. W-CDMA became the basis for the 3G GSM standard (later named the Universal Mobile Telecommunications System, or UMTS for short). UMTS evolved over the years to become UMTS HSPA+. European and Asian network operators would purchase large 60-80MHz chunks of spectrum in auctions designed similarly to the one done by the FCC to operate UMTS. This Euro/Asian-centric band planning was kept in mind when DoCoMo developed W-CDMA, and when the first few releases of the UMTS standard came out after the initial release. It wasn’t until much later, with release 5, that they started working on optimizing for smaller blocks.
Fast forward to 2003, the FCC realized that there was a problem when they were preparing Section 1 of the Advanced Wireless Services band (AWS-1 band for short) for auction. They wanted to design a new band that was as close to the European bands as possible, but the spectrum was currently being used by satellite cable and military operations. Unfortunately, it was the only band they could take spectrum from, since the planning stages for the switchover to digital TV hadn’t even been finalized yet! And since satellite cable was failing in the market, the military remained the only problem. By 2006, they worked out a band block plan (similar to the band block plan for PCS) and a framework to migrate the military off with the cooperation of the Department of Defense. In 2005, they announced the auction that would be taking place in 2006. Deutsche Telekom (having owned T-Mobile USA for only five years now) announced that they would be buying nationwide licenses for AWS-1 in the auction. Needless to say, they did. Unfortunately for them, they didn’t get all of what they wanted. In several regions, they only acquired 10MHz of spectrum, particularly in areas that Verizon and U.S. Cellular outbid them.
In any case, T-Mobile USA and Deutsche Telekom immediately began soliciting vendors to prepare AWS equipment for their 3G network in 2007. By 2008, the first AWS 3G devices arrived and network deployment began. Unfortunately for all of us, the economic downturn hit us hard and slowed down T-Mobile’s deployment of AWS 3G. They made up for this in 2009 by deploying the very latest 3G technology, HSPA 3.6 and then upgrading it to HSPA 7.2 later in the year.
Yet another quick note: In late 2008, the 3rd Generation Partnership Project (the group that handles standardization of new UMTS technology) approved the initial standards for 3GPP Long Term Evolution for the Universal Mobile Telecommunications System, or UMTS LTE for short, for UMTS Release 8 alongside single carrier and dual-carrier HSPA+. UMTS HSPA+ and UMTS LTE are improved and standardized together from this point onward.
In 2010, T-Mobile began deploying single carrier HSPA+ across the nation, and in 2011 deployed dual carrier HSPA+. They took advantage of their late-to-the-game status and managed to get highly efficient equipment for much cheaper, with support for easily bolting on LTE. While in the planning stages for the deployment plans for HSPA+, it was discovered that their AWS assets would not be sufficient for nationwide deployment at the highest level. However, they did discover that with judicious reshuffling of network banding and refarming of their PCS spectrum, they could deploy at least dual-carrier HSPA+ across the entire nation, with some markets having quad-carrier HSPA+. More importantly, they could also deploy LTE in areas where they are weak spectrally and aggregate the connection with HSPA+ to boost speeds even further.
So T-Mobile developed a plan: they would deploy as far as they could with AWS for HSPA+, and when it came time to refarm PCS for 4G, they would deploy HSPA+ on PCS zones where they had 30MHz or more spectrum and LTE on unused 10MHz AWS zones. If AWS was currently in use, they would deploy HSPA+ on PCS and leave room for LTE and GSM. They could do other configurations as needed. The added benefit was that if they prepared for it right, they could reduce their overall costs by a huge percentage!
At this point in time, T-Mobile is just now refarming PCS spectrum. In a majority of the markets where T-Mobile has both PCS and AWS spectrum with no AWS used, it is quite likely that they will deploy HSPA+ only on PCS because most of the remaining markets only have 10MHz of AWS and 30MHz or more of PCS. AWS will likely be reserved for LTE in these areas. In areas where they’ve got plenty of AWS spectrum and they’re using it for HSPA+, they’ll dedicate about 10MHz of PCS to GSM and 10MHz to HSPA+. If they have 20MHz of PCS or less in a market with lots of AWS, there will be no deployment of HSPA+ on the band, but there will be scaling back of GSM to 10MHz to prepare for LTE.
T-Mobile will deploy LTE. They have to eventually. The main reason for it is because they need the spectral optimization that LTE offers. When LTE was being designed, they kept in mind that American cellular network operators were desperate for a new, more optimized system that could handle their tighter spectral constraints. They made it possible for LTE to deployed in 1.4MHz, 5MHz, 10MHz, and 20MHz band configurations. Even in 5MHz deployment, LTE can offer comparable average speeds to dual-carrier HSPA+ with the appropriate backhaul and efficient core network.
Another reason for it is network aggregation. LTE will eventually support aggregating HSPA+ and LTE together to make faster, more efficient connections to the network and to the Internet. With aggregating HSPA+ and LTE together, they could make it possible to get 100Mbps with a latency of 60ms on average on a mobile device without LTE-Advanced. With deploying LTE-Advanced, they could go even further!
Finally, each LTE cell can support three to four times as many heavy usage subscribers as an HSPA cell. That is a substantial capacity improvement over HSPA+. This is because LTE is an OFDMA based technology, and explicitly uses MIMO to carry more connections at once. If T-Mobile plans to support the entire country with their network, they’ll need the capacity gains assuming each person has one smartphone, one tablet, and one hotspot connected to five devices. For 300 million people, that would require a lot of capacity. While their HSPA+ network is capable of supporting everyone, there would be some speed decreases and latency increases. A hybrid LTE/HSPA+ network promises to actually offer higher speeds than HSPA+ alone, lower latencies than HSPA+ alone, and support even more people with their current spectral assets.
Right now, they are playing the waiting game. Doing so is smart, because the costs for deploying the equipment needed to support LTE on AWS and PCS are quickly coming down with MetroPCS, Leap Wireless, AT&T, and eventually Verizon Wireless deploying on that band. Additionally, if they do partner with DISH Network to use their S-band spectrum, they’ll want to make sure the equipment they buy supports it.
Finally, we have to note that T-Mobile USA has been preparing for LTE on their core network and backend infrastructure for over two years now. Their UMA implementation became the basis of the VoLGA standard that Deutsche Telekom wanted to use for all T-Mobile LTE networks across the world. However, none of the other operators wanted to use the technology, so Deutsche Telekom relented and switched their support to the more expensive IMS based technology. Consequently, T-Mobile USA deployed all the necessary components to run IMS on their core network and made a new Wi-Fi Calling solution that uses it. As far as we know, they are the first in the world to commercially deploy IMS for voice, SMS, MMS, and other circuit-switched services. It would be trivial for T-Mobile to change the client software to make it run over LTE or HSPA+. By preparing all the core network and backend infrastructure for LTE this far ahead, their nationwide LTE deployment costs will much lower.
I seriously doubt it will cost $9 billion unless Deutsche Telekom wants to buy more spectrum to use for LTE. For now, though, they don’t really need to. As long as they are smart about how they use their spectrum, they can hold off acquiring more spectrum for a few more years.