2G GSM was one of the most successful things to ever come out of Europe's technology sector. So when the next generation came around, everyone bought into the wireless hype. Billions of dollars spent buying spectrum, which led to a big crash and then a slow 3G rollout.
Everyone wondered how the telecoms could ever make back the billions spent on a 3G investment. What could ever be big enough? In this video, we relive the 3G wars, the 3G crash, and the Jesus phone that brought them all back to life. ## Beginnings 0G wireless networks existed, but I am not going to talk about them. The first commercially available 1G wireless networks date to the 1980s. In the United States, Bell Labs launched a trial 1G cell network in Chicago in 1977. Then they
and Motorola launched the Advanced Mobile Phone System or AMPS in 1982. 1G deserves to be labeled as such - not only because they were the first commercial deployments - but also because they established the tenets of the modern cellular industry as we know it. What are those tenets? First, the concept of a telecom operator buying a license for a band of spectrum from the local government authority. It basically begins here.
Second, the telecom does not try to cover a massive service area with just one massive tower. Instead, they set up many smaller base stations that only have to cover a single cell area. The third and final tenet of modern cellular systems introduced by AMPS and other 1G networks was seamless handoff/access. When someone exits a cell and enters another, any in-progress calls are seamlessly transferred between the two. ## FDMA AMPS did this by splitting the whole licensed band into non-overlapping slices. Each base station in a cell gets its own slice for communicating with handsets in its area.
No two neighboring base stations get the same frequency slice, so to prevent interference. Imagine the spectrum as like a piece of land. This makes sense. Like land, spectrum is a limited resource. When the station connects to the handset during a phone call, the base station is essentially building a road on that land. But an AMPS base station can't communicate with two or more handsets using the same slice of spectrum. To deal with this, the base station cuts its own slice up into yet smaller subdivisions and gives each person in the cell their own sub-divided slice.
So going back to the metaphor, that means one road to one driver. This idea is known as Frequency-Division Multiple Access or FDMA. If the metaphor works, then you can see the drawbacks of the approach. Since we need to space out the roads to keep the drivers from potentially driving into one another, we can't have that many roads on our land. 1G calls sounded bad. The handsets were too big and suffered poor battery life. And the handoff between towers was unstable. Other than that,
everything was awesome. By 1990, there were 11 million wireless subscribers. ## 2G 2G networks introduced named standards. Managed by working groups and organizations, standards help commercialize basic cellular technologies. They offer telecoms an ecosystem that saves them money through economies of scale while also establishing international connectivity norms. Wireless generations can accommodate multiple standards. The most widely adopted 2G standard
was Europe's GSM. GSM originally began in 1983 as the Groupe Spécial Mobile committee, working on a new standard after some too-expensive 1G deployments in Europe. GSM was notable for its use of Time-Division Multiple Access, or TDMA to let the same spectrum carry more data bits. What is TDMA? As I mentioned, the 1G pioneer AMPS network used Frequency-Division, where we split up the land into many little pieces. If FDMA is like building a road for each user in the cell, then TDMA is like those vacation house timeshares or an AirBnb for a digital nomad. Someone visits a house for a weekend or so, and then they jump to another house next door after an allotted time period is up.
With TDMA, we slice up the spectrum like as with FDMA. But each handset - instead of getting their whole own slice all the time - only gets their slice for a certain time slot. Time guard bands are left in between each handset's transmission to prevent interference. Coordinating this round-robin switching behavior can get complicated. But the upside is that we can enable more users per channel slice and squeeze more capacity by making the time guard bands smaller. The first 2G GSM networks went up in Europe in 1991, with commercial operations beginning in 1992. The first non-European telecom to sign up was Telstra.
GSM broke out. In part because of its simplicity, wealth of convenient features like smart cards and SMS texting, as well as being first to market. By 1994, over 100 telecoms had signed up including several in the US, which GSM entered when it added the 1900 megahertz band.
GSM eventually captured 80% market share of the 2G market, sweeping aside America's cdmaOne - remember that name - and Japan's Personal Digital Cellular standard. ## The 3G Controversy As 2G GSM was being finalized - the GSM ecosystem players started pondering what should come next. What eventually emerged was the trend towards better internet access. While
GSM did initially have that, the data speeds were not great - like 9.6 kilobits per second. To put in perspective, to download a 3 megabyte MP3 file with 9.6 kilobits per second, that would take like 44 minutes. A few years before GSM launched, the European Union began to define a new global standard called Universal Mobile Telecommunications System or UMTS. UMTS would not only handle calls but also offer a stepwise improvement in data transfer rates. But throughout the 1990s, it started to dawn on everyone that mobile was a world-changing technology. By 2005, there would be 1.2 billion wireless subscribers using GSM
technology. And the Europeans sat at the very heart of the new revolution. That not only conferred immense global prestige, but also financially benefitted Europe's domestic wireless companies like Nokia and Ericsson. After building networks at home, they exported to developing countries leapfrogging wired networks directly to cellular. Like, Eastern Europe, Southeast Asia and mainland China. The last of which would become the world's biggest GSM market. Moreover, 3G was initially marketed early on in the 1990s to the telecoms as an entirely new thing. The telecoms were not super thrilled about that considering that many of them
had just finished building their 2G networks and needed to make back their investments. Thus grew a view within the European body then in charge of GSM - European Telecommunications Standards Institute or ETSI - that creating a new standard was not necessary. Rather, they should either built an evolutionary 3G onramp or just extend GSM to handle higher data rates. So as the Europeans built up UMTS, research insights from that work flowed into GSM. The first data extension was something called General Packet Radio
Service or GPRS. You can probably call this a 2.5G wireless technology. GPRS raised GSM data rates to 53.6 and then later 114 kilobits per second. Which was good enough to let people send emails and even kind of browse the web in the form of Wireless Application Protocol, WAP. After GPRS, we had the Enhanced Data Rates for GSM Evolution or EDGE, which offered speeds of up to 384 kilobits per second using the same fundamental GSM technology under the hood. GPRS and EDGE were good extensions for GSM. However their limitations also demonstrated the family's fundamental shortcomings.
Using the TDMA method - as in, the timeshare - would always prevent GSM from matching the higher data transfer rates that another technology can do. ## CDMA So now, we should pick back up a latent thread - CDMA. CDMA stands for Code-Division Multiple Access. And the name gives you a hint as to how this accursed thingy works. With TDMA, we still had to divide the spectrum same as FDMA, leaving guard bands in between the divisions. Like as before, this was not efficient. Can we not?
With CDMA, you can. Each transmitting CDMA user is given a unique codeword and is allowed to broadcast using the whole spectrum band. In other words, everyone can talk at the same time but because they are essentially speaking different "languages" it works. Qualcomm introduced the first CDMA networks with the cdmaOne standard back in the 2G days. It did fine in the United States, but the Europeans opted not to adopt it for the GSM standard due to its complexity. And that probably was the right choice. Though GSM blew ahead in terms of worldwide subscribers, cdmaOne subscribers grew quite well into the late 1990s.
The CDMA concept was once again presented for inclusion into UMTS. But for whatever reason - most likely a bad case of "Not Invented Here" syndrome - ETSI struggled to adopt it. A faction in Europe instead backed an enhanced version of TDMA, Advanced TDMA, and this went back and forth. ## No Global 3G For 3G, the goal had always been to collaborate to create a global standard. That was the initial dream of the United Nations' International Telecommunications Union or ITU. But GSM's unexpected success changed the dynamics. Europe wanted a single global standard, but one strongly influenced by them, just like with GSM.
The Americans felt that they had to keep any standard from being entirely dominated by Europe. GSM winning the 2G era led to many American wireless companies losing business. And American consumers looked upon with jealousy at their European friends' international roaming powers. And Japan too needed to get into the mix, because their own 2G national standard PDC left them with a sort of Galapagos Island effect that hurt their companies' competitiveness. So in late 1997, the ITU chose not to back a single standard but a family of them. All somewhat-compatible and called "3G". Each had their own backers racing for adoption by 2000. Ergo why they were named the IMT-2000 family. That same year, ETSI decided to adopt an air interface standard known as WCDMA out of five candidates for their UMTS system. Developed as a Japan-Europe collaboration,
WCDMA is a CDMA-based technology known for its efficient use of spectrum. Japan's biggest wireless telecom NTT DoCoMo preferred WCDMA because they had limited amounts of spectrum. And since it had been a joint Japan-Europe effort, the Europeans felt that WCDMA was enough of theirs to move forward. I need to note that WCDMA is not exactly the same as UMTS. WCDMA specifically refers to the
air interface - which was how the handset communicates with the base station. But to keep it simple, I shall refer to the two - UTMS and WCDMA - as just WCDMA. Anyway to promote WCDMA, in 1998, ETSI joined with several other standards bodies in the US, Japan, Korea, and China to create the Third Generation Partnership Project, or 3GPP. ETSI and the other standards bodies handed over WCDMA and the ongoing GSM data extension work, meaning GPRS/EDGE etc, over to 3GPP. ## The Patent Fight Over in the US, the rather litigious Qualcomm soon came to believe that ETSI - working in cahoots with Ericsson and Nokia - was suppressing the Qualcomm flavor of CDMA behind the scenes.
Ericsson and Qualcomm were also then in the midst of a great patent infringement battle dating back to 1996. In 1998, Qualcomm announced that they would withhold licenses for their patents - despite widely held norms that they should do so - so long as “politics” continued to play a role in 3G standards development. American politicians backed them. Qualcomm then sponsored the formation of a new competing standards group in late 1998. This group - confusingly named 3GPP2 - continued developing the cdmaOne standard as a North American one, creating CDMA2000.
Now WCDMA and CDMA2000 aren't all that dissimilar, the differences are largely superficial. For example they both depend on CDMA, which led to problems because Qualcomm held a lot of key patents in CDMA. So by being unwilling to license, they were essentially holding the 3G ecosystem hostage. This dragged on until Qualcomm and Ericsson finally settled their lawsuits in March 1999 with a broad cross-licensing patent agreement. It came ahead of
a deadline set by the ITU, which found the whole thing appalling. But Ericsson more likely chose to settle and embrace CDMA because they realized that Qualcomm was about to hit the titanic mainland Chinese market. Business newspapers reported that the Chinese government told the United States that they would open up certain key markets if the US backed their ascension to the World Trade Organization. This included dividing the old Chinese phone monopoly into three companies - China Mobile, China Telecom and China Unicom - and then having each of the new companies adopt a standard.
China Telecom used the American CDMA2000 standard. This was seen as a major coup. ## Crazy Spectrum Outside of the patent fights, the early 2000s saw very slow 3G adoption. Maybe we can blame it on there being two major but incompatible 3G standards. But as I mentioned, the standards aren't all that different. And Nokia and Qualcomm eventually released chipsets supporting both standards. So there were other issues at play. There might have been too much hype over a potential 3G rollout - perhaps tied in to the larger dotcom and fiber boom going on at the time. Anticipating a great mobile bonanza ahead,
the European governments charged the telecoms a great deal of money for 3G spectrum. In April 2000, the British government held a 3G spectrum auction that raised some $30 billion from six telecom bidders. Note that this is just the spectrum, we haven't even considered the cost of building the actual network.
But people at the time were thinking that the 3G industry would be worth $200 billion. Considering that, $5 billion for a single license made sense right? More troublingly, the British government pushed the telecoms to pay all upfront for their licenses. The telecoms had no choice but to borrow from the financial markets to make those payments, which had severe consequences down the line. The Germans and French launched similarly massive auctions. Germany sold a set of
licenses for a staggering $45 billion. One went for $7.7 billion alone, won by a joint venture of partners including Hong Kong-based Hutchison Telecom, run by the cunning Li Ka-shing. Shortly after that, Hutchison sold their 50% stake in that JV, saying that the price was too high considering that they would be just one of six different telecoms in the German market. Li told reporters: > The prospect of 3G (third-generation mobile phone networks) is very good but anything, even the best, should have a price in the commercial society ... I don't believe the license price can go up endlessly Li Ka-shing should know. His empire includes a telecom in
the super-crowded Hong Kong market. When he sells, everyone should pay attention. Over in France, France Telecom also pulled out from the bidding after the French government set a fixed price of $5 billion for their 3G spectrum. Later auctions in the Netherlands, Switzerland, Italy, and Poland also turned out to be disappointments - reflecting the impact of the crashing telecommunications bubble and the deflating of 3G hype. And then when the 3G networks finally went live, they disappointed consumers. People were promised FAST data speeds of 2 megabits per second, but that turned out to be theoretical. The best thing
available in 1999 was EDGE, which only got to 384 kilobits per second. So the customers stayed away. ## Where is the Money? Sure the industry players - from the telecoms to the banks issuing debt for them - can upgrade their networks. But what would be the thing or service to get more people to sign up for these pricey 3G services? Nobody knew at the time.
The best value-added data service that early adopters like the telecom 3 in the United Kingdom could come up was "video telephony". But that required both sides to have a more advanced handset. Other telecoms focused on offering business services like corporate VPNs, advertising, and facilitating transactions. Industry analysts at the time projected advertising revenue growing from $600 million in 2000 to $31 billion in 2010, 51 times over. Unrealistic by any measure. One operator joked that 3G stood for "Games, Girls and Gaming".
Another joke went that UMTS really stood for "Unproven Market, Technology and Services". Jokes aside. At the very heart of it, 3G needed a killer app. Without it, all you were going to get from the telecoms were vague "statements of intent" but no real commitments.
## Then Came the iPhone And then came the iPhone, the first of a class of smartphones with large screens, full web browsers, and the compute power of a mini-computer. The first iPhone was a 2G device - presumably for battery life reasons. And in the United States, launched exclusively on the AT&T mobile carrier. It sold fairly well - taking about 74 days to hit a million devices sold. Even taking into account how restricted the old 2G iPhone was, users still ate 15 times more data than before, and 50% more than what AT&T internally projected. But the real boom came when Apple released its second-generation iPhone 3G and opened up its App Store with third-party apps. The iPhone became the center of its users' lives,
used to check email, send messages, watch unsavory video, and more. Plus AT&T started subsidizing the $599 handset if they signed up for a two-year contract. After subsidy it costs just $199. Nice. We also had the rise of Android, which rose in the iPhone's wake. And since Android was
supported by several manufacturers and was not locked to a single carrier, these grew even faster than the Apple device. ## Data Crunch Adoption, data consumption, and strain on the networks all accelerated. iPhone users consumed 24 times more data than the average. Their old-school unlimited data plans - at just $30 a month, no less! - did not help either. iPhone usage grew data traffic on the AT&T mobile network by 5,000% from 2006 to 2009. From 2007 to 2010, AT&T mobile data volume grew by 8,000%. And competing US telecom Verizon, a notable backer of Android, found that Android users also consumed several times more data than ordinary phone users.
Locked in a race with Google to capture market share, Apple was in no mood to artificially restrain its world-changing product to accommodate AT&T's pleading requests to help its network. AT&T saw no choice but to invest billions. From 2007 to 2010, they invested a total of $37 billion into network upgrades. Nevertheless, the ATTFAIL hashtag continued to circulate on Twitter. ## Conclusion The iPhone turned out to be the Killer App that everyone in the telecom industry had been waiting for. It got so bad, that analysts were out there predicting a "data crunch" apocalypse looming ahead in 2013. A time when the wireless networks might grind to a halt - rendering all the beautiful smartphones useless.
The wireless industry folks fought for a glorious wireless future for 3G. But when that future finally came, it was nothing like what they thought it would be. The industry scrambled to adapt, with lasting effects on the coming 4G wireless era.
2025-02-13 00:26