Well, my mother didn't make me smart enough so I understand all this electronics stuff... but what I get from the article is that new technology will enable future phones to download porn faster. Yay !
That is true, I think for most Anantech reader where we could understand a High Level details of Node, Fabs, ISA, x86 / ARM, Decoder, Pipeline, GPU Compute etc, Wireless Spectrum, transceiver, Modem and Baseband are something we have hard of but know not much about it. I admit i had to look up a lot of things just to get along with this article when a lot of these wont previously explained in Anand's article at all.
Would be great if Anand could do a Basic explanation 101.
So, first, that initial slide showing the Modem SS block and mod/demod with links to VPE and QDSP6 is painful. Even at a high level that is so inaccurate it hurts.
Second, 4 layer MIMO in r9 is lip service. The necessary spec information to support a 4 Rx configuration is missing, r9 effectively caps out at 4x2 2layers, which is being actively deployed in Korea right now.
Finally, the Cat3 limitation is not on the usable number of Resource Blocks (RBs) but on the number of information bits (uncoded bits) that can be transmitted in a single transmission instance (subframe) and the number of soft bits that are stored in the coded buffer for rate matching. 9615 and 8960 devices can use all 100RBs available in 20MHz, provided the amount of information to be sent is kept within the limit on information bits. This prevents the highest coderates from being used with that many RBs, but there is no limitation on how many (or few) RBs that can be allocated.
That powerpoint presentation is generally full of twisted non-information and about the same level of detail and accuracy as a wrinkle cream ad. But that's exactly what it is, an ad.
Obviously parts of that are just marketing, but this is a first step to getting them to disclose a much more comprehensive high level block diagram. Also I'd love to know how you'd draw the diagram as well.
Stepping away from the more sexy discussion points of a smart-phone is always going to be tough, yet you have done so. With LTE, as you mention, being far from a simple "is this phone LTE capable," specifications such as ports/bands are in greater need of proliferation.
I have a question, if the carriers are refarming the existing bands for LTE, what is stopping a device which is LTE capable using that band, if it is available in the 2G/3G spectrum? One doesn't require different filters for different modulations, do they?
Now if only I could crack open one of these base stations...
I've enjoyed listening to the podcasts and hope to have this article assimilated before the CES deluge. (Humor me. January and I'm feeling aspirational...)
Many months ago in a conference in Atlanta where the CTO was present I remember hearing about if how the Metro merger goes through and once their spectrum is consolidated with ours that in areas where the have 40 Hz available we will be deploying 148 Mpbs; though I do not expect handsets that can support those speeds to be available as we light up the first few cities. How quickly the merger goes through will determine that as much as our ability to deploy it though.
"The last real remaining black box from my point of view is the cellular connectivity side of things."
There is the 4GAmerica's site which has a large number of very up-to-date and extraordinarily good white papers on the technical details of past and future GSM releases. That's one way to keep up but, like you say, it doesn't tell us what the chip vendors (or the telcos) are doing.
I expect Qualcomm would not answer this, but is their willingness to start being a little more open a fear of Rosepoint in the future? Intel is presumably headed for their turf, and that must scare the bejesus out of them... I wouldn't be surprised if, as Rosepoint becomes more real, Qualcomm starts becoming a lot more flexible about just how it's willing to sell off its IP, allowing you, if you want, to buy blocks you can stick on your SOC (rather than Qualcomm branded chips), and even designing custom such blocks for you if you provide the specs on exactly what you want.
I second the question: What Qualcomm thinks of Intel full digital radio approach? Another question, they added Beidou support, what about European Galileo system?
Beidou has been operational in China since 2011, and across the asia pacific region since last month; Galileo isn't expected to reach initial operating capacity until the middle of the decade. Until last October there weren't even enough sats in orbit to begin testing it; so while Qualcom is probably working on something they couldn't've put something known to work in the current silicon. Presumably when it's closer to IOC in a year or three they'll add support for it. Until then it'd be a waste of silicon for devices that will largely be retired before it's usable.
The WTR1605 looks like a baby step that didn't really move anything forward, unless you want to get into the Chinese market a little bit.
It really has no advantage for the US market with the single mid band diversity port since the major carriers use both the low and mid bands, while the small carriers have only mid band spectrum.
This is way off topic I think, but does anyone know what they use for staking on that board? The layer looks to be perfectly around each SMT component near the chip, and I am curious what they use.
After reading the article I have a better understanding of why Anand hired you. I am not surprised that Qualcomm wants to service multiple markets with a single chip so incremental improvements to increase the number of markets the chip may address is typical market leader behavior. I do not know whether a sidebar or pod cast is the best vehicle, but a rant providing some compare and contrast between the Qualcomm baseband strategy vs the Intel strategy would be helpful. Also, it would be a good place to incorporate your preliminary read on how the digital radio stuff Intel recently demonstrated may affect the marketplace and the block diagrams.
You must go into the weeds to establish credibility. However, once you got it, few care about the weeds. Everybody wants more red meat in the ARM vs Intel thing and this is a lovely opportunity to serve some up. Kindly share the benefit of your insight.
Interesting, i was late to the article and there is only 10 comments, compared to 3 page on ARM vs Atom. People not interested in LTE / Baseband at all??
Anyway, so do LTE UE 4 offer better bandwidth efficiency then UE 3? Since both only required 20Mhz, but UE 4 gives up to 150Mbps.
Apart from Beidou and TDS-CDMA, WTR1605 seems like a small step, no size reduction?
Are there any power improvement with 9x25? LTE seems to be draining battery a lot.
So i gathered all current Qualcomm already support VoLTE, we are only waiting for carrier to support it, right? And may be off topic, why aren't carrier doing it / Faster?
I am not sure if i am right, the more port there are, the more supported band / wireless spec there will be. It seems to be one of the reason iPhone 5 could not come with world wide LTE supported. So wouldn't ditching GSM help? ( 4G is here.... time to ditch 2G right? )
How do WiFi and Bluetooth fits into the scenario? They are all wireless tech, why do they requires another chip? Couldn't Qualcomm fits those in?
I remembered there was a article about Intel Digital Radio. I admit i still do not understand much of it. Any relation to this? Or is Digital Radio more on the antenna side of things.
VoLTE increases the load on their 4g networks by reducing the load on 2g/3g; since 4g is only going to get more crowded with time while 2g is becoming a ghost town and 3g will become one in the next few years as LTE deployments are completed, VoLTE does nothing beneficial for the carriers in the short term.
Long term it's needed to let them shut down their legacy networks; but that's at least 4-5 years out according to occasional talking points they make (and if Sprint/iDEN is any indication even farther out in the real world) which makes it not worth enabling for handsets that will probably be junked well before it happens.
You forgot one point: Efficiency also matters for Voice, as long as the network is robust enough. In terms of efficiency: LTE is by far more spectral efficient than 2G. GSM for example needs 5 to 7 frequencies to operate one network, UMTS and LTE only 1. The carrier modulation and coding of LTE is 1.5 to 2 times more efficient than UMTS. In terms of robustness: LTE is as robust as GSM from a modulation and coding point of view (something UMTS is not). That means the same bandwidth voice codec can run on much less spectrum on an LTE network compared to GSM, while still being almost as robust against interference. In consequence, VoLTE should be very desirable for the carriers. Even in the short term. They should be able to re-farm some of their 2G frequencies even faster in case the speed up the VoLTE introduction.
Carriers have the harder part of implementing VoLTE. Since no carrier has a strictly 4G network, and generally have a larger 3G footprint than 4G, the voice traffic has to be able to be dynamically switched between the 3G and 4G networks. While this is happening for data, voice handovers are less forgiving. While the equipment to do this is all available now, to integrate it into the existing network with the same level of service requires integration testing and bug fixing.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
20 Comments
Back to Article
wavetrex - Friday, January 4, 2013 - link
Well, my mother didn't make me smart enough so I understand all this electronics stuff... but what I get from the article is that new technology will enable future phones to download porn faster. Yay !iwod - Sunday, January 6, 2013 - link
That is true, I think for most Anantech reader where we could understand a High Level details of Node, Fabs, ISA, x86 / ARM, Decoder, Pipeline, GPU Compute etc, Wireless Spectrum, transceiver, Modem and Baseband are something we have hard of but know not much about it. I admit i had to look up a lot of things just to get along with this article when a lot of these wont previously explained in Anand's article at all.Would be great if Anand could do a Basic explanation 101.
ImSpartacus - Tuesday, January 15, 2013 - link
That's what I love about Anandtech. Anand does spectacular explanatory articles and the other writers aren't too shabby.Belegost - Saturday, January 5, 2013 - link
So, first, that initial slide showing the Modem SS block and mod/demod with links to VPE and QDSP6 is painful. Even at a high level that is so inaccurate it hurts.Second, 4 layer MIMO in r9 is lip service. The necessary spec information to support a 4 Rx configuration is missing, r9 effectively caps out at 4x2 2layers, which is being actively deployed in Korea right now.
Finally, the Cat3 limitation is not on the usable number of Resource Blocks (RBs) but on the number of information bits (uncoded bits) that can be transmitted in a single transmission instance (subframe) and the number of soft bits that are stored in the coded buffer for rate matching. 9615 and 8960 devices can use all 100RBs available in 20MHz, provided the amount of information to be sent is kept within the limit on information bits. This prevents the highest coderates from being used with that many RBs, but there is no limitation on how many (or few) RBs that can be allocated.
Wwhat - Sunday, January 6, 2013 - link
That powerpoint presentation is generally full of twisted non-information and about the same level of detail and accuracy as a wrinkle cream ad.But that's exactly what it is, an ad.
Brian Klug - Monday, January 14, 2013 - link
Obviously parts of that are just marketing, but this is a first step to getting them to disclose a much more comprehensive high level block diagram. Also I'd love to know how you'd draw the diagram as well.-Brian
Kogies - Saturday, January 5, 2013 - link
Stepping away from the more sexy discussion points of a smart-phone is always going to be tough, yet you have done so. With LTE, as you mention, being far from a simple "is this phone LTE capable," specifications such as ports/bands are in greater need of proliferation.I have a question, if the carriers are refarming the existing bands for LTE, what is stopping a device which is LTE capable using that band, if it is available in the 2G/3G spectrum? One doesn't require different filters for different modulations, do they?
Now if only I could crack open one of these base stations...
watersb - Saturday, January 5, 2013 - link
I've enjoyed listening to the podcasts and hope to have this article assimilated before the CES deluge. (Humor me. January and I'm feeling aspirational...)Brilliant stuff.
jaysns - Saturday, January 5, 2013 - link
Many months ago in a conference in Atlanta where the CTO was present I remember hearing about if how the Metro merger goes through and once their spectrum is consolidated with ours that in areas where the have 40 Hz available we will be deploying 148 Mpbs; though I do not expect handsets that can support those speeds to be available as we light up the first few cities. How quickly the merger goes through will determine that as much as our ability to deploy it though.Drazick - Saturday, January 5, 2013 - link
Will You, Please, Update Your Google+ Page?It would be much easier to follow you.
name99 - Saturday, January 5, 2013 - link
"The last real remaining black box from my point of view is the cellular connectivity side of things."There is the 4GAmerica's site which has a large number of very up-to-date and extraordinarily good white papers on the technical details of past and future GSM releases. That's one way to keep up but, like you say, it doesn't tell us what the chip vendors (or the telcos) are doing.
I expect Qualcomm would not answer this, but is their willingness to start being a little more open a fear of Rosepoint in the future? Intel is presumably headed for their turf, and that must scare the bejesus out of them...
I wouldn't be surprised if, as Rosepoint becomes more real, Qualcomm starts becoming a lot more flexible about just how it's willing to sell off its IP, allowing you, if you want, to buy blocks you can stick on your SOC (rather than Qualcomm branded chips), and even designing custom such blocks for you if you provide the specs on exactly what you want.
SydneyBlue120d - Sunday, January 6, 2013 - link
I second the question: What Qualcomm thinks of Intel full digital radio approach?Another question, they added Beidou support, what about European Galileo system?
DanNeely - Sunday, January 6, 2013 - link
Beidou has been operational in China since 2011, and across the asia pacific region since last month; Galileo isn't expected to reach initial operating capacity until the middle of the decade. Until last October there weren't even enough sats in orbit to begin testing it; so while Qualcom is probably working on something they couldn't've put something known to work in the current silicon. Presumably when it's closer to IOC in a year or three they'll add support for it. Until then it'd be a waste of silicon for devices that will largely be retired before it's usable.toyotabedzrock - Saturday, January 5, 2013 - link
The WTR1605 looks like a baby step that didn't really move anything forward, unless you want to get into the Chinese market a little bit.It really has no advantage for the US market with the single mid band diversity port since the major carriers use both the low and mid bands, while the small carriers have only mid band spectrum.
ishbuggy - Saturday, January 5, 2013 - link
This is way off topic I think, but does anyone know what they use for staking on that board? The layer looks to be perfectly around each SMT component near the chip, and I am curious what they use.dealcorn - Sunday, January 6, 2013 - link
After reading the article I have a better understanding of why Anand hired you. I am not surprised that Qualcomm wants to service multiple markets with a single chip so incremental improvements to increase the number of markets the chip may address is typical market leader behavior. I do not know whether a sidebar or pod cast is the best vehicle, but a rant providing some compare and contrast between the Qualcomm baseband strategy vs the Intel strategy would be helpful. Also, it would be a good place to incorporate your preliminary read on how the digital radio stuff Intel recently demonstrated may affect the marketplace and the block diagrams.You must go into the weeds to establish credibility. However, once you got it, few care about the weeds. Everybody wants more red meat in the ARM vs Intel thing and this is a lovely opportunity to serve some up. Kindly share the benefit of your insight.
iwod - Sunday, January 6, 2013 - link
Interesting, i was late to the article and there is only 10 comments, compared to 3 page on ARM vs Atom. People not interested in LTE / Baseband at all??Anyway, so do LTE UE 4 offer better bandwidth efficiency then UE 3? Since both only required 20Mhz, but UE 4 gives up to 150Mbps.
Apart from Beidou and TDS-CDMA, WTR1605 seems like a small step, no size reduction?
Are there any power improvement with 9x25? LTE seems to be draining battery a lot.
So i gathered all current Qualcomm already support VoLTE, we are only waiting for carrier to support it, right? And may be off topic, why aren't carrier doing it / Faster?
I am not sure if i am right, the more port there are, the more supported band / wireless spec there will be. It seems to be one of the reason iPhone 5 could not come with world wide LTE supported. So wouldn't ditching GSM help? ( 4G is here.... time to ditch 2G right? )
How do WiFi and Bluetooth fits into the scenario? They are all wireless tech, why do they requires another chip? Couldn't Qualcomm fits those in?
I remembered there was a article about Intel Digital Radio. I admit i still do not understand much of it. Any relation to this? Or is Digital Radio more on the antenna side of things.
DanNeely - Sunday, January 6, 2013 - link
VoLTE increases the load on their 4g networks by reducing the load on 2g/3g; since 4g is only going to get more crowded with time while 2g is becoming a ghost town and 3g will become one in the next few years as LTE deployments are completed, VoLTE does nothing beneficial for the carriers in the short term.Long term it's needed to let them shut down their legacy networks; but that's at least 4-5 years out according to occasional talking points they make (and if Sprint/iDEN is any indication even farther out in the real world) which makes it not worth enabling for handsets that will probably be junked well before it happens.
thm82 - Monday, January 7, 2013 - link
You forgot one point: Efficiency also matters for Voice, as long as the network is robust enough.In terms of efficiency: LTE is by far more spectral efficient than 2G. GSM for example needs 5 to 7 frequencies to operate one network, UMTS and LTE only 1. The carrier modulation and coding of LTE is 1.5 to 2 times more efficient than UMTS.
In terms of robustness: LTE is as robust as GSM from a modulation and coding point of view (something UMTS is not).
That means the same bandwidth voice codec can run on much less spectrum on an LTE network compared to GSM, while still being almost as robust against interference. In consequence, VoLTE should be very desirable for the carriers. Even in the short term. They should be able to re-farm some of their 2G frequencies even faster in case the speed up the VoLTE introduction.
jhh - Tuesday, January 8, 2013 - link
Carriers have the harder part of implementing VoLTE. Since no carrier has a strictly 4G network, and generally have a larger 3G footprint than 4G, the voice traffic has to be able to be dynamically switched between the 3G and 4G networks. While this is happening for data, voice handovers are less forgiving. While the equipment to do this is all available now, to integrate it into the existing network with the same level of service requires integration testing and bug fixing.