This page is an archive from our previous website. Please check out our new website where you can read new COMMENTARY eNewsletters, TELL IT LIKE IT IS blog posts or Press Releases.
 
In the meantime, questions are already being asked about the how much life is left in 2G and 3G systems.
 
Commentary

Before 700 MHz (The Fate of 2G and 3G Systems)

Monday, April 21, 2008
 

I have written enough about 700-MHz for a while, but I will be staying involved in the public/private sector debate about the D block and the new auction at some point this year. We know who has won the spectrum rights across the country and who has won pockets of the spectrum. The networks wont be built for a while, and the final clearing date for the TV stations is Feb 17, 2009, if that really happens. At the moment, there are many questions concerning how quickly consumers will transition from their analog-only TVs to digital sets, not only in their main TV viewing area but for the second, third and fourth TVs spread around their homes. Meanwhile, serious planning is taking place around the United States and the infrastructure vendors, tower companies and others are looking forward to a couple of great years with the 700-MHz and AWS build-outs.

 

In the meantime, questions are already being asked about the how much life is left in 2G and 3G systems. The short answer is, A lot! Voice pays the bills and 2G systems (GSM and CDMA) both work well for voice. Today, most 3G systems are only about data services. On the CDMA side, EV-DO Revs 0 and A are data-only technologies (with push-to-talk coming soon, and VoIP at some point). On the GSM/UMTS side, as far as I know, most of the mixed GSM/UMTS systems are using GSM for voice and UMTS mostly for data with steps down to EDGE and then GPRS where there is no UMTS coverage.

 

The mixing of voice and data on the same carrier has always been problematic. In the early days, when we started running data over voice systems for the first responders there were all kinds of problems with voice dispatch delays, additional noise in the vehicles from the data services and the loss of voice capacity. As time progressed, these same issues came to commercial wireless. CDMA2000 1X mixes voice and 120-Kbps data on the same carrier. If it is managed well, no one sees much of a difference, but in times of high or peak voice usage, data rates are lower because less of the carrier is available for data.

 

Some say that when we get to VoIP over 3G and 4G systems, it wont matter because a bit is just a bit. However, that is not necessarily the case. The packets that will be transporting VoIP are much more time sensitive than the data packets and will, therefore, have to have priority. These packets must arrive at the other end in a certain order and within a small time window in order for the voice to be put back together and to sound even remotely like a real voice. Quality of Service (QoS) is needed to make voice over IP work properly with either 3G or 4G systems and my bet is that it will be some time (2012 or 2013) before we see widespread use of VoIP over the wide-area wireless broadband networks.

 

In the meantime, 3G networks will continue to be expanded, not only in the United States, but everywhere in the world. Second-generation networks will not go away any time soon. For the most part, they are paid for, and their coverage is still better than 3G networks in most parts of the world. Voice, which still accounts for >70% of most networks revenue, is still the most important part of a network. There is a drive to increase data services and you might look at the latest numbers as reported by RCR and think that data usage is climbing nicely. For example, AT&Ts Average Revenue Per User (ARPU) at the end of the 4th Quarter was $50.26 and its data ARPU was $9 or just about 18% of its overall revenue. Verizons ARPU was $51.46 with data being $11.06 of that, or just over 21%. Others show similar figures, but these numbers reflect ALL data including SMS, MMS and 3G broadband data. In Verizons case, it might also include ARPU from its MediaFLO mobile TV offering, which is off-network, but it is hard to tell.

 

Yes, data numbers are climbing each quarter, but if you remove SMS and MMS from them, the actual 3G data income numbers, while still growing, are not growing as fast as the network operators would like them to. A lot of time, money and attention is being paid to how to increase 3G revenue numbers and, just as important, how to get some of the voice-only customers to try high-speed data, or even a single application that would be of some value to them. Once customers are familiar with how data services work, and especially if they are receiving one data service they really like, chances are they will experiment more with other data services and applications.

 

Location-based services including turn-by-turn directions, locating ATMs and other such activities, will help grow data usage as will games, especially interactive and multi-player games, video downloads and business applications. Interaction with YouTube, Facebook and MySpace all help drive the youth market to the world of 3G, but if we are to entice those above that age group there has to be something compelling for them to want to access. Is it the Internet via a browser? Is it sports for a segment of the population? Mobile TV? In reality, it is different things for different groups of customers and even for individual customers.

 

Growing 3G wireless data services reminds me of the early days of the BlackBerry, only this time it is more difficult because it is not about a single application, rather, it is about access to diverse applications and services. When the BlackBerry first came out, it was easy to determine if there was an interestall you had to do was identify those who lived on email and whose email was as important as their voice mail. If you showed people a BlackBerry and they said, So what?, you knew they did not value email enough to want to carry it around on their belt or in their purse, and their eyes usually glazed over. But today it is not about showing off or discussing a single application, it is about finding that one application or service someone would really like to have with them all of the time. To some, it is still email (not really a 3G application), but to some it is an application that works better over 3G and it is these types of applications that need to be promoted to the (so far) voice-only crowd.

 

Some applications can benefit from 3G speeds but dont necessarily need them. For example, many payment programs for wireless banking, bill paying and the like, which are fast becoming a hot market, dont necessarily need the speed of 3G, but as the applications become more graphical, having a 3G connection will make a difference. The industry is struggling with how to get voice customers hooked on 3G data services. There are thousands of developers working on slick applications delivering all kinds of information. In fact, one of the problems is that in some cases we might have too many applications or services, making them difficult to find. This is one reason so much time and effort is being spent on new search tools, engines and even voice searchto help us find what we want, when we want it.

 

But we are also on the eve of other types of devices that consume 3G bandwidth: appliances or consumer products that are wirelessly enabled. My Amazon Kindle is one of the first of these. For those not familiar with it, the Kindle is an electronic book and I can order and download books from the Kindle store at Amazon using the Kindle itself or via the website. If I am using the Kindle directly, I am communicating with the Amazon site over the Sprint EV-DO network. If I buy a book on the website it is sent to me via the Sprint network and it only takes a few minutes to arrive.

 

To my thinking, this is one of the models of the future. I bought my Kindle so I own it outright, I dont have a data subscription with Sprint nor do I pay Sprint a penny for the service. Instead, I purchase a book from Amazon for $9.95 and it arrives on my KindleAmazon pays Sprint for use of the network. Think about the number of wireless devices in use today (somewhere around 3 billion) and then think about how many appliances such as the Kindle, dash-mounted navigation systems, game boys, portable DVD players and other devices there are in the world where a wireless chipset could be embedded to add real value to the device and the information it contains. Granted, many of these devices will not use a lot of data, more bursty data such as my Kindle, but they will use wireless data over and over again and increase the networks overall data ARPU.

 

Possibly the biggest market will be for machine-to-machine communications. Today, companies such as Aeris provide the ability for machines to stay in contact with each other and to call home when they have data to report or need to be fixed (remember the refrigerator add on TV?). Today, they are using short bursts of datasmall amounts that are sent over a control channel or as SMSbut they will begin using 3G data services so they will be able to send longer streams of data or include graphics with what they send, and provide even better services for those who own these machines and need to know their status.

 

Then there is the use of 3G and 4G services for video feeds from surveillance cameras, command-and-control of our homes lights, garage doors and other things, the use of movie trailers to entice us to attend the local theater and watch a first-run movie and a hundred other things. They are beginning to come now and will continue to increase at a fast clip.

 

I think we can provide broadband to much of rural America using EV-DO and UMTS and eventually 4G services. If there is an economic model for rural America, I believe it will be to use existing commercial networks for mobile and fixed voice and data services. The load for each cell site will be light enough that a home in rural America will be able to make use of an EV-DO or UMTS connection and have access to the Internet a lot less expensively than trying to build out a 2.5-GHz WiMAX system or using satellite access, which is done today but with a lot of latency.

 

So 2G will stay around for a long time for voice, 3G will be around even longer, first for data and then for data and VoIP, and 4G will be deployed where demand for data is highest and 3G networks are running near capacity. Eventually, 4G systems will take hold and the others will be phased out, but there has to be a return on investment for the network operators, and junking one network for another has to make economic sense. Yes, it costs the network operator less to deliver a kilobyte of data over 3G than over 2G, and it will cost even less on a 4G network that is IP end-to-end, but that is only one reason to move forward as technologies are introduced.

 

Analog voice, the most spectrum-inefficient of all of the wireless technologies, had a life span of about 27 years. TDMA did not last as long, but that was because the decision was made to replace one 2G technology with another. Will 3G survive for 27 years? Will it take 10 years before 4G is the prevalent technology deployed around the world? I am not sure anyone has the answers yet as part of this will depend on the demand for data services, and the demand for data services depends on many different things and groups and hooking people on the capabilities afforded by broadband wide-area wireless data services.

 

Andrew M.Seybold

COMMENTS: This is an archived post. Commenting is no longer available.

Syed Hosain - 04/22/2008 13:32:18

Hi, Andy!

I would like to thank you very much for the very positive mention of Aeris in your Commentary! One thing I would say is that we have *already* added the services that you predict we will do in the future. As someone once said: "The Future Is Now".

In CDMA, we provide Digital Control-Channel short data service (MicroBurst), high-performance high-reliability SMS, as well as 1XRTT and EVDO wireless data and voice. For applications that need the higher data throughput capacity, or voice, on an "on-demand" basis, along with the immediacy of event notifications enabled by MicroBurst or SMSDirect.

These services are fully deployed in North America (including Canada and Mexico) with a transparency, and managed network performance, that is quite unmatched for these kinds of applications!

Thanks again!

Syed Zaeem Hosain,
CTO, Aeris Communications, Inc.

David Boettger - 04/22/2008 14:25:14

How wonderfully ironic that carriers globally embarked on a mad dash to build out 3G without the semblance of a plausible business case and even now -- EIGHT YEARS LATER -- everyone is still scratching his head, trying to figure out how to fill the bandwidth. It's especially ironic that AT&T had the benefit of seeing how poorly UMTS was doing in terms of increasing ARPU elsewhere in the world and STILL chose to build it. Talk about destroying shareholder value.

And now it's happening again with "4G" (whatever that is). Rather than trying to plan out how to make money off of LTE, carriers once again are in a mad-dash to out-technologize each other. Newsflash: LTE isn't enough better than HSPA to justify deploying it! Newsflash #2: Subscribers don't give a damn if they're using 2G, 3G, or 4G, as long as the application is usable! One would think that carriers would know this about their subscribers by now.

If there were any real competition in cellular (there isn't), then I daresay that we'd have a lot more creativity on the application front and a lot less focus on G-ness and who's deployed the latest revision of the technology. Which is why the outcome of the 700 MHz auction so depressing (though completely predictable).

Indi Vidual - 04/22/2008 19:16:07

Reaction from the dilletante owner of a New Zealand WiFi service (Zenbu):
http://siliconinvestor.advfn.com/readmsg.aspx?msgid=24522954

===================


Andrew joins the old telecoms measure of arpu as significant, which it is not. Arpu is a useless measure. What matters is profit per dollar invested, not revenue per user. To get arpu increasing, all they have to do is kick off the low paying customers and keep those who will pay more for the service.

They could keep doing that until they have one customer with an arpu of $10,000 per month. They would have the best arpu on Earth and they would not be in business for long.

Another company might have 5 billion subscribers with an arpu of $1 a year. It's a good bet that the $1 arpu per year company would be doing better than the $10,000 arpu per month company. Arpu is one of the dopiest ideas I have seen. But they all seem to go on about it.

<In the meantime, 3G networks will continue to be expanded, not only in the United States, but everywhere in the world. Second-generation networks will not go away any time soon. For the most part, they are paid for, and their coverage is still better than 3G networks in most parts of the world. > Being "paid for" isn't a factor in whether a technology is obsolete or not. All that matters is whether a new technology can be brought in by a competitor and do better than the incumbents. 2GHz 3G is too expensive for regular humans. That's why it isn't used.

Andrew goes on about finding the secret sauce single application which will get people onto the 3G bandwagon. All that's needed is for the people running the thing to open a big fat low cost pipe and get out of the way.

I don't use 3G. I would like to have 3G available, but the greedy service providers want $1 a megabyte on an arpu-increasing minium usage, use it or lose it, "plan". Telecom wants $8 a megabyte for casual users. They are insane. Normal humans won't pay that. Vodafone isn't much better. I don't have a "plan" at petrol stations. I ride my little motor scooter into whichever I like once a month or once a year and buy $5 or so. I don't have to buy $80 a month whether I want it or not. Nor do I buy petrol by the minute - I buy it by the litre. I don't pull in and say "I'll have 3 minutes of petrol please". I pay for what I take delivery of.

Zenbu, an amazing new wireless service in New Zealand charges only 10c a megabyte and there is no "plan". http;//www.Zenbu.net.nz People buy the amount of credit they want [$5 minimum] and use it when they like. They can use 1c a month or 1c a year. I don't even know what the arpu is. Mostly people buy a $10 credit. But some buy $15, $20 and a few people buy credit in $100 chunks. To some people, $100 is neither here nor there and they don't want to goof around buying credit frequently. They have a big fuel tank on their vehicle and they say "fill 'er up" - they can drive from here to London and back again on a single tank. Others just use the 5 megabyte free credit for a new account.

Telecom has a Wi-Fi service too. They charge $10 an hour for Go-'s sake. So somebody touring NZ decides to check their email to see if Mum has written about Dad's operation. They buy a $10 an hour credit. Damn, there's no email. They fiddle around looking at this and that which they are not really wanting to do because their friends want to get going and checks again at the end of the hour. Still no email. So they have to pay another $10 a few hours later to check again. That's great arpu, and totally rotten service.

Zenbu enables them to check for email for NO charge. 5 megabytes is enough to check for an email. If they want to write an email, and use some megabytes doing this and that, maybe banking, checking accommodation, events, look for ATMs, get a map of where they are, find a BP service station, they can buy $10 credit which doesn't expire, and take all the time they like. They can come back 40 times in the next 3 days waiting for that email and it'll cost them almost nothing. That's very low arpu and totally great service.

The reason Zenbu was started was because Telecom and Vodafone and the service providers around the world have been so greedy that Wi-Fi has become the default technology. Telecom should be able to do it cheaper than Zenbu, but not with wall to wall executives in multi-storey buildings on huge salaries harassing subscribers with arpu-boosting, tricky, confusing and annoying expensive marketing gimmicks costing a fortune on television and in print in costly stores selling phones out of glass cases.

Fire the people, cut the price, open the pipe, make it fast. That's how to get data moving.

All applications are 3G applications. Email included. A millisecond is a millisecond. Our telomeres are wearing out and every millisecond that goes by while we dither is another millisecond down the gurgler. 2G is not good enough.

More telecom oligopoly thinking: <So 2G will stay around for a long time for voice, 3G will be around even longer, first for data and then for data and VoIP, and 4G will be deployed where demand for data is highest and 3G networks are running near capacity. Eventually, 4G systems will take hold and the others will be phased out, but there has to be a return on investment for the network operators, and junking one network for another has to make economic sense. Yes, it costs the network operator less to deliver a kilobyte of data over 3G than over 2G, and it will cost even less on a 4G network that is IP end-to-end, but that is only one reason to move forward as technologies are introduced. >

It is irrelevant whether a network has been profitable, or even had its cost returned. What matters is the best way of doing things. If 4G is a lot better and the 2G and 3G networks have not made money, too bad. They are obsolete. If there's a competitor, they won't care whether the greedy incumbents have made money on their legacy networks. Zenbu doesn't care that Telecom's Wi-Fi and EV-DO systems have not made profits. We care about Happy Customers.

OFDM in 450MHz looks like a good bet. A LOT better than 3G in 2GHz.

3G is easy. Open the pipes, make it cheap, get out of the way. Forget about arpu. Don't even measure it.

Andrew Seybold - 04/23/2008 11:12:55

Great to have more comments--Idai--yours is an interesting post--and obviously biased against what you consider the "bad guys" who have provided telecommunicatinos services for years. But in order for my readers to understand everything about you operation we would need to know the other facts assocated wtih the network.

How much of the City/Country is covered?
Does the system work inside?
What was the total Capital expenditure to build the network?
How many access points per square KM?
How many do you add every month because of interference?
How many back-haul links per square KM? What is there data speed and how many access points do they cover?
What is your monthly Opex? Sites, Backhaul, office,insurance, everything?
Are you making any money or just chruning cash?

By the way CDMA EV-DO works great at 450 MHz and yes, it will better than anything at 2.5 GHz!

Thanks for the post

Andy