RupertBainesThere is a bit of debate on 4G (or "faux G" as Dan Jones of Light Reading nicely put it). Many people have been using the term "4G" to refer to WiMAX and LTE for some time with both Sprint (Clearwire) and Verizon using it heavily in their promotion. As I'll explain below, I think that makes sense, but this has recently become controversial for two reason (intriguingly opposite ones). First, the ITU announced that LTE-Advanced and WiMAX 2 (802.16m) are candidate technologies for IMT-Advanced and thus meet its definition of "4G". Secondly, T-Mobile has responded to the other carriers adverts by starting its own 4G - using HSPA+ which most people would probably describe as 3.5G.
I have seen this categoric statement:  "LTE-Advanced (and its counterpart WiMAX 2 or 802.16m) are the first 4G air-interfaces, as they meet the ITU requirement of 4G: 1Gbps stationery and 100MBps mobile.  

I don't agree.

While ITU certainly control the definitions of IMT-Advanced (and before that IMT-2000 which included WCDMA and WiMAX) I am not sure they are the sole arbiter of 4G-ness. "4G" is not a standard or a defined thing. IEEE creates 802.16 standards, WiMAX Forum certifies it as WiMAX. 3GPP creates LTE standards. ITU decides what makes the IMT-Advanced cut. But the term "4G" is owned by none of those bodies - it is the consensus of the industry which will define it. ITU certinaly don't own a trademark on the term 4G, or didn't even invent the terminology of 1G, 2G or 3G.

Indeed, I think a definition based on data-rate is fundamentally misconceived. The point of the G is to describe something meaningful, to give a taxonomy or a structure which helps understanding. Data-rate is weak proxy to that.

As such, the ‘G’ of generation refers to architecture change. 1G was analog; 2G was TDMA narrowband; 3G was CDMA with a lot of circuit-switched architecural features – and 4G is OFDMA, with very wideband (20MHz), and a ‘flat architecture’ with end-to-end IP. Each of those technology changes was fundamentally different to the one that went before. With that perspective, LTE is 4G - it is radically different to what went before (CDMA to OFMDA, circuit-switched-ish to all-IP).

That is a fundamental difference: Something changes between generations. But mixing WiMAX in as 3G, and trying to say LTE is generationally different from LTE-Advanced just gets silly.

Within each of these you may get iterations or updates; for example, from GSM to GPRS, from WCDMA to HSPA - or from LTE to LTE-Advanced. Sometimes people use intermediates (GPS is often referred to as 2.5G, with EDGE as 2.75G; HSPA as 3.5G). LTE-Advanced is not so different to LTE.  Indeed, there are few obvious changes - you have to be pretty detailed to even describe the differences between them. To describe it as "advanced" is probably a marketing over-statement, but to be safe lets call it 4.5G.

Incidentally, those of you with long memories may recall similar debates about ten years ago; was CDMA2000 3G or not. Tim Kridel helpfully linked to this snapshot of the past, with a discussion on whether or not 1X was a 3G standard. 

Phil Solis of ABI has written on this several times, and I think he is absolutely right.

Data-rate is a peculiar way to assess things. First, because it is very articificial - what you see in the real world is loosely related to the headline rates of a standard and is very variable. Second, because the standards evolve - as noted above there are things like 2.75G. By choosing an arbitrary data rate the ITU has ended up with an arbitrary transition. Although technology architecture is related to data-rate, it is inexact: it would be analagous to saying that human generation is defined by age. I have a friend who is older than his aunt (big Irish family: grandfather had 12 children with a 25 year span, and the eldest was married and had a son before the youngest daughter was born) - but we do not say "she is younger, she can't be his aunt because that is a different generation". Similarly, Enhanced EDGE (the latest update of GSM) can go faster than the standard variety of TD-SCDMA (pre-HSPA). Does that meant they are no longer 2.xG and 3G - or does it mean that data-rate is not a meaningful way of defining generations?

Some people have pointed to spectrum as a defining marker, but that is misleading too. Although early bands are better suited to older technologies (primarily because they are narrower) there is no specific, causal, relationship. Refarming is the most explicit disproof of the idea, or the fact that in different countries different generations use the same bands.

No, the simple, logical taxonomy is that generation means what is says; it is a significant, discrete thing, which reflects a fundamental change to architecture and air-interface.   WiMAX and LTE are 4G.