Apple Posts iOS 5.1 Final: Build 9B179 - Update: "3G" Logo changed to "4G" on AT&T

19 Comments

View All Comments

• name99 - Wednesday, March 7, 2012 - link

  Let me give you an example of what I mean, and example that I think is quite realistic.
  
  Suppose a new telco starts up with an innovative business model. They're going to use the flexibility inherent in OFDM to create a cellphone system using scavenged frequency. Maybe they find 1.5MHz free between two other uses, or maybe they use a wider band (5 or 10MHz) but notch out a large number of the frequencies because there are existing narrowband users scattered throughout the band. Either way, their effective bandwidth let's say, after pilots and notches is 1.5MHz.
  
  They run LTE on this, so how well can they do? We start with with 64QAM, so that's 6bits/Hz. Use 5/6FEC so we're down to 5bits/Hz. Use 2x2:2 MIMO, and we're at 10bits/Hz. So best case scenario, we can run our PHY at 15Mbps.
  
  So, the question is: should this service advertise itself as 4G?After all, it IS USING 4G technology --- the whole thing that made it possible was OFDM and the ability to notch out certain frequencies.
  But the peak speed available for download is 15Mbps --- basically exactly the same as the peak for HSPA Rel5.
  
  So what is the value in insisting that one be labelled 4G because it uses OFDM, while the other be labelled 3G because it is based on WCDMA? That's not a useful distinction for users. The distinction that matters for users is the speed, and the speeds are essentially the same.
  Yes, the latency of the LTE system is probably lower --- but not necessarily dramatically so. Best case LTE is a lot better than worst case HSPA, but bad LTE can be comparable to good HSPA.
  
  Think about this example. Like I said, what I see here is a whole lot of fetishizing of something that is incidental (whether the fundamental "robustness to delayed echoes" mechanism of the transmission is CDMA or whether it is OFDM) over what actually matters to most users which is bandwidth.

• brybir - Thursday, March 8, 2012 - link

  Well, we all appreciate the technical articles but you are missing the authors point.
  
  Most end users do not care AT ALL about the intricacies of building and maintaining cellular networks and devices.
  
  What we do care about is the relative abuse of marketing terms that end up leaving the end user disappointed.
  
  I bought a T-Mobile Android phone a few months ago so that I could take advantage of the higher data cap plans and to upgrade from my "3G" speed to "4G" speed. Both AT&T and T-mobile were using the exact same terms, so logically I suspected that even if there were small differences in the actual outcome, they would be reasonably comparable. As you know, T-mobile's "4G" is nothing like AT&T 4G when on an LTE network. So, I was left somewhat disappointed in that regard. This is what he is talking about...companies marketing the crap out of a common term when the actual implementations are VASTLY different. It seems and feels misleading and deceptive.
  
  Perhaps in the future the phone companies will start marketing like the internet companies...you know..."Come to T-Mobile and get out fastest service yet, our 14.4 Mb/sec plan with a 4GB data cap" rather than the current system.

• name99 - Thursday, March 8, 2012 - link

  So you're exactly agreeing with my point. Customers perceive 4G as a speed term, not as a technology term, and HSPA rel7 is fast enough that it should be binned in the 4G category not the 3G category.

• name99 - Wednesday, March 7, 2012 - link

  "
  So on the comment down below, I want to point out that spectral efficiency of GSM/EDGE is not what I'm getting at, at all, actually.
  "
  
  Then what did you mean. You agree with my point re Rx Diversity, I assume? So, what did you mean?

• Brian Klug - Wednesday, March 7, 2012 - link

  I think we're actually in total agreement here, as far as HSPA+ goes on the ever-subjective "G" scale (which of course doesn't mean anything) it is definitely faster than WCDMA 384 kbps which is the original place the 3G line was drawn (along with 1xRTT as well). Obviously further 3GPP releases make WCDMA orders of magnitude faster and deserve recognition.
  
  To some extent the solution there is really the Nokia/Android route, eg. showing either 3.5G, or simply treating the end user like an adult and showing H or HSPA+, or (what I'd really like) just the straight up UE category and feature set that are currently working. I'm completely in agreement there.
  
  The real issue here is one of AT&T-specific carrier incursion into iOS merely for the sake of furthering AT&T's advertising. I don't think at this point that HSDPA 14.4 (one 5 MHz carrier, 16QAM) is arguable as 4G. Quite honestly I draw the line for when it's more or less sane to call HSPA+ "4G" or not at DC-HSPA+ as we've spelled out a few times now, but again that's an entirely subjective definition that I've imposed.
  
  The irony of course is that AT&T isn't running 64QAM or DC-HSPA+ at all. It's single carrier 16QAM everywhere right now.
  
  As for the GSM/EDGE thing, I've learned that this is now enabled on the GSM/UMTS iPhone 4Ses on every carrier but AT&T, so it's clearly just a spectrum refarming thing. What I was alluding to however is the fact that the phone has no Rx diversity at all except for when you're on 1x/EVDO/WCDMA air interfaces, and thus the phone does exhibit attenuation behavior just like the 4 in those cases.
  
  -Brian

• name99 - Wednesday, March 7, 2012 - link

  "What I was alluding to however is the fact that the phone has no Rx diversity at all except for when you're on 1x/EVDO/WCDMA air interfaces, and thus the phone does exhibit attenuation behavior just like the 4 in those cases."
  
  What makes you say this?
  Engineering-wise, it's perfectly possible to make a GSM/EDGE phone use Rx diversity, just like it was possible (and some base stations did) use Rx diversity for 802.11g.
  Obviously Apple has dual cell-frequency antennas.
  
  So this boils down to a claim that the phone makes no attempt (not even something really trivial like switch-and-stay every 20ms) when it's connected to GSM/EDGE. This is obviously possible --- Qualcomm (which I assume is the chip making these decisions) could simply have omitted that functionality from their chip. Though, that is supposed to be a world-ready chip, and if EDGE Evolved is taking off in some countries (it was tested in India and China in 2009) it would be make sense to have Rx diversity in the chip.
  
  So, to reiterate, what makes you believe that there is no Rx diversity in iPhone 4S (or any other dual antenna 3GPP phone)? This is something that is tricky to test given that we're dealing with EM field variations over a length scale of cm and a time scale of tens of ms.
  
  As for the rest, well, I've said my piece, you've said yours. It seems that you are willing to agree that what's most useful to users is a speed indication, and 3G vs 4G serves that.
  
  Beyond that
  - ATT only: I have no interest in this either way. It is possible that ATT actually has plans for fairly widespread deployment of 64-QAM and/or 2x2:2 HSPA MIMO in the near future, and this is the essential point that made Apple feel it was worth making the change.
  It's also possible that other carriers that actually support HSPA (perhaps at a certain release or feature level) also get this. Do we have enough REAL evidence (ie people who know for a fact that their network is running a decent level of HSPA) , presumably in Europe, maybe Japan, to really know yet?
  
  - 3G vs H vs H+ vs 4G indicators: Apple doesn't provide nerdy info to its users. That's just the way they roll. I have pointed out (in other disagreements I have had with you) the meaninglessness of obsessing over "number of bars" and its geek cousin "dB level" --- the only indication that really matters is an indication of the speed one can expect. But Apple has never gone in for indicators of network speed, or disk transfer speed, or CPU utilization level, or amount of swapping, or anything else of that sort. Even on a Mac you have to install 3rd party apps if you want those sorts of status indicators.
  
  I'm not defending this practice --- I have iStat Menus installed on my macs, and, if I had the choice, I'd love to have similar functionality in my iPhone and iPad status bars. But this is the choice Apple has made --- very granular indications --- basically what people expect and nothing more --- and they're unlikely to change from that to an indication that "you are currently connected to a tower that supports the following feature set.... Your phone negotiated the following options.... You are connected using these frequencies... Over the past three minutes here's a histogram of the goodputs you saw, the modulation + FEC that was used, etc etc"
  It would be kickass if that info were available, but, like I said, that's just not the way Apple does things.

• ltcommanderdata - Wednesday, March 7, 2012 - link

  Any chance for quick benchmarks to see if there are performance improvements for mobile Safari and GLBenchmark?

• Andrew.a.cunningham - Wednesday, March 7, 2012 - link

  A real quick, unscientific SunSpider .91 run on my 4S running 5.1 yields a score almost identical to the number under 5.0 in our 4S review. Looks like Safari wasn't a focus this time around.

• name99 - Wednesday, March 7, 2012 - link

  "There's another big contributing factor for why the 3G toggle has been re-removed as well on the iPhone 4S - lack of Rx diversity on GSM/EDGE. "
  
  I'm not sure what you are trying to say here.
  Rx diversity is not a property of the modulation system or the tower electronics or anything else. Rx diversity is a way to combat the fast fading that occurs in an environment with many reflections. All it requires is that the receiver have smarts (which can vary from minimal to very sophisticated) to sense the signal quality at both antennas and either choose the better antenna or (smarter) combine the signals from both antennas. It is different from concepts like multi-stream MIMO or STBC which do rely on an interaction between the tower and the receiver.
  
  I suspect what you REALLY MEAN is that GSM/EDGE (as deployed by ATT) is substantially less spectrally efficient than HSPA and LTE, and that is the reason ATT is trying to reduce its use.
  
  [For those who care:
  Even this is not completely correct --- it is possible to substantially increase the spectral efficiency of EDGE by using the same bag of tricks as used by HSPA and LTE --- higher order modulation, better FEC algorithms, MIMO. This collection of ideas makes up EDGE Evolved.
  ATT has decided not to bother with this route (and they are correct to do so --- the one big trick that LTE has that neither EDGE nor HSPA has is OFDM. and that's a trick worth switching to for a variety of reasons).
  
  The larger point is that these things are not as cut-and-dry as people seem to think. This stuff is not magic --- there is a specific, and finite, set of ideas that are applied to make digital wireless work and then work better. Depending on particular concerns, one may be most interested in spectral efficiency or in the peak bandwidth available to a handset, but either way, talking about "2G" or "3G" or" LTE" without talking about a particular technology and why it is relevant to one's point is not useful.]