On the road to 4G

Depending on how you look at it, it's already here

A hot topic as last week's Interop conference was 4G wireless. What, 4G already? Do we even have 3G at this point? And just what is 4G, anyway?

Before we get too far ahead of ourselves, it's important to keep in mind that mobility is now a core driver of the evolution of networking overall. It's pretty clear that many people are now moving toward using a cell phone as their primary or even their only phone. Farpoint Group estimates that 15% of cell phone users already fall into these categories, and we will move toward 40% over the next 10 years. Cell phones are also moving toward 70% to 80% penetration of the total population of the industrialized world. In other words, the voice market is saturating.

Both of these eventualities have been key drivers in the development and deployment of 3G cellular, which began more than 10 years ago in Europe as the Future Public Land Mobile Telecommunications System (FPLMTS), later the IMT-2000 (International Mobile Telecommunications-2000, and then 2000 MHz.) and today UMTS (Universal Mobile Telecommunications System). UMTS is one of two primary 3G systems, the other being CDMA2000. 3G is a category of all-digital cellular systems, with an emphasis on broadband, primarily for data, but also for increased voice capacity and improved operations and management for carriers deploying 3G.

1G, by the way, was analog, and 2G was digital, but narrowband. 3G was initially defined entirely by (broadband) data throughput specifications: 144Kbit/sec. when moving at high speed, 384Kbit/sec. when moving relatively slowly, and up to 2Mbit/sec. when stationary or indoors. Rate of motion turns out to have a dramatic impact on effective throughput due to various forms of fading as well as Doppler shift. And while the above data rates are peak speeds (i.e., available under ideal conditions on the downlink from the cell to the user only), they are a dramatic improvement over the circuit-switched peak of 14.4Kbit/sec. on 2G networks. And 3G is in fact now widely available throughout the world.

Today, we are looking at deployments with greater than 2Mbit/sec. peaks, such as 1X EV-DO Revision A, which peaks at 3.1Mbit/sec., and HSDPA, which can range up to 14.4Mbit/sec. We often call these 3.5G systems, since 4G requires a little more, and not necessarily in terms of throughput.

And this is where it gets difficult, because, for example, like "broadband," 4G is an imprecise term. There are two schools of thought here. The first is to take the obvious path and simply define 4G as faster than 3G or 3.5G. One example is the UTRAN LTE (UMTS Terrestrial Radio Access Network Long-Term Evolution) project now under way within the Third Generation Partnership Project, which develops functional specifications for UMTS. The design points for UTRAN LTE are performance of 100Mbit/sec. downlink and 50Mbit/sec. uplink peak throughput. Mobile WiMax could have peak performance of around 75Mbit/sec., as could 1X EV-DO Revision B. These speeds are obtained via a combination of multiple-input, multiple-output, lots of bandwidth (20MHz channels at a minimum vs. today's 1.25-5MHz), more sophisticated modulation and advanced antennas. Throwing technology at this opportunity isn't a problem.

But devoting that kind of throughput to a single user is. Given the expense involved in further upgrading cellular networks to these 4G technologies (including new handsets), and especially the cost of spectrum, and I don't think we'll see anything like this anytime soon. Service will simply be too expensive. And contention among multiple users on any given channel will likely limit throughput to 3Mbit/sec. to 5Mbit/sec., regardless.

There is another definition of 4G, however, and that's a mobile, broadband, all-IP service with support for time-bounded traffic. This implies VoIP, as well as high data throughput again on the order of 3Mbit/sec. to 5Mbit/sec. (effective data rate, as opposed to peak). The problem here would be in converting the SS7-based cellular network to IP, and that won't happen for a very long time. I estimate 15 years at least.

In the meantime, however, consider that a metro-scale Wi-Fi network easily fits the definition of 4G. Consider: Such deployments are indeed all IP, their coverage area is potentially indeed metro-scale, they provide very high throughput to users (5Mbit/sec. to 10Mbit/sec., and up to 20Mbit/sec. are certainly possible, even on fairly well-subscribed systems, especially with the availability of 802.11n), and they can support VoIP and even streaming video. And unlike cellular, bandwidth and throughput are symmetrical with some variance of course due to differences in antennas and transmit power. But, in one sense anyway, we have 4G today. And as I expect metro-scale Wi-Fi deployments to accelerate, the need for 4G cellular may be moderated, so to speak, at least a little bit.

Copyright © 2006 IDG Communications, Inc.

7 inconvenient truths about the hybrid work trend
Shop Tech Products at Amazon