Does a MIMO receiver have a whole analog front end for each antenna? Or can you use one PLL, ADC and Filter, and switch between the antennae?
MIMO receiver...
Started by ●November 9, 2006
Reply by ●November 9, 20062006-11-09
regarding this, processing has to be seperated for each antenna in terms of ADC and Filter, this is what i have seen as based on a prototype built by one company. regards particlereddy porterboy76@yahoo.com wrote:> Does a MIMO receiver have a whole analog front end for each antenna? > Or can you use one PLL, ADC and Filter, and switch between the antennae?
Reply by ●November 9, 20062006-11-09
On 9 Nov 2006 05:27:23 -0800, porterboy76@yahoo.com wrote:>Does a MIMO receiver have a whole analog front end for each antenna? >Or can you use one PLL, ADC and Filter, and switch between the antennae?Switching between antennas only achieves switch diversity, which is good but not nearly as good as most MIMO algorithms. Most people don't really call a switched system MIMO, just switched diversity. e.g., 802.11b systems have had switch diversity in the receiver since day one, but nobody calls them MIMO. For the most part, yes, the receivers are separate. This is because most of the effective combining/decoding algorithms are digital, so each stream must be digitized separately. It would probably be possible to build something like an MRC receiver that is analog so that the combining is done before the ADC, but that would preclude other algorithms from being used. Space-time coding (other than Alamouti) or spatial multiplexing will definitely require multiple antennas in (typically) both the transmitter and receiver. Eric Jacobsen Minister of Algorithms, Intel Corp. My opinions may not be Intel's opinions. http://www.ericjacobsen.org
Reply by ●November 9, 20062006-11-09
Eric Jacobsen <eric.jacobsen@ieee.org> writes:> On 9 Nov 2006 05:27:23 -0800, porterboy76@yahoo.com wrote: > >>Does a MIMO receiver have a whole analog front end for each antenna? >>Or can you use one PLL, ADC and Filter, and switch between the antennae? > > Switching between antennas only achieves switch diversity, which is > good but not nearly as good as most MIMO algorithms. Most people > don't really call a switched system MIMO, just switched diversity. > e.g., 802.11b systems have had switch diversity in the receiver since > day one, but nobody calls them MIMO. > > For the most part, yes, the receivers are separate. This is because > most of the effective combining/decoding algorithms are digital, so > each stream must be digitized separately. > > It would probably be possible to build something like an MRC receiver > that is analog so that the combining is done before the ADC, but that > would preclude other algorithms from being used. > > Space-time coding (other than Alamouti) or spatial multiplexing will > definitely require multiple antennas in (typically) both the > transmitter and receiver.Eric/porterboy76, If you consider the analog signal bandlimited to B, then isn't it true at least in theory that if you analog sample the two inputs each at least 2*B you could then multiplex the digital portion? As long as the analog sampler meets Nyquist, you wouldn't lose anything, i.e., you could use those same digital algorithms once you get the signals into the digital domain. Am I missing something? -- % Randy Yates % "Rollin' and riding and slippin' and %% Fuquay-Varina, NC % sliding, it's magic." %%% 919-577-9882 % %%%% <yates@ieee.org> % 'Living' Thing', *A New World Record*, ELO http://home.earthlink.net/~yatescr
Reply by ●November 9, 20062006-11-09
On Thu, 09 Nov 2006 17:13:14 GMT, Randy Yates <yates@ieee.org> wrote:>Eric Jacobsen <eric.jacobsen@ieee.org> writes: > >> On 9 Nov 2006 05:27:23 -0800, porterboy76@yahoo.com wrote: >> >>>Does a MIMO receiver have a whole analog front end for each antenna? >>>Or can you use one PLL, ADC and Filter, and switch between the antennae? >> >> Switching between antennas only achieves switch diversity, which is >> good but not nearly as good as most MIMO algorithms. Most people >> don't really call a switched system MIMO, just switched diversity. >> e.g., 802.11b systems have had switch diversity in the receiver since >> day one, but nobody calls them MIMO. >> >> For the most part, yes, the receivers are separate. This is because >> most of the effective combining/decoding algorithms are digital, so >> each stream must be digitized separately. >> >> It would probably be possible to build something like an MRC receiver >> that is analog so that the combining is done before the ADC, but that >> would preclude other algorithms from being used. >> >> Space-time coding (other than Alamouti) or spatial multiplexing will >> definitely require multiple antennas in (typically) both the >> transmitter and receiver. > >Eric/porterboy76, > >If you consider the analog signal bandlimited to B, then isn't it >true at least in theory that if you analog sample the two >inputs each at least 2*B you could then multiplex the digital >portion? As long as the analog sampler meets Nyquist, you wouldn't >lose anything, i.e., you could use those same digital algorithms >once you get the signals into the digital domain. Am I missing >something?Randy, That's a good idea, but often M>2, so it'll get worse as you add antennas. Certainly there's nothing to stop anyone from doing it this way, but the switch will add distortion/loss/etc. If the cost tradeoff to one ADC with M*fs vs M ADCs at fs works, then why not? Eric Jacobsen Minister of Algorithms, Intel Corp. My opinions may not be Intel's opinions. http://www.ericjacobsen.org
Reply by ●November 10, 20062006-11-10
> If you consider the analog signal bandlimited to B, then isn't it > true at least in theory that if you analog sample the two > inputs each at least 2*B you could then multiplex the digital > portion? As long as the analog sampler meets Nyquist, you wouldn't > lose anything, i.e., you could use those same digital algorithms > once you get the signals into the digital domain. Am I missing > something?Well, what about synchronisation? If you have a different AFE on each of antennae, you have N different PLLs, N different ADCs and N different AGCs. Digitally you are then trying to combine N different signals which have different sampling phase/frequency offsets, different carrier phase/frequency offsets, different symbol timing and different nominal gains. In paticular there may be large differences in the group delays, which means you would need digital buffering. Is this not exteremely suboptimal (or at least very difficult to compensate for). In fact, I guess I am making a lot of assumptions which may just be untrue, since in general I havent a clue what happens before the signal gets digitized... especially in MIMO. I'm a definite noob when it comes to AFE stuff!
Reply by ●November 10, 20062006-11-10
On 10 Nov 2006 01:03:34 -0800, "porterboy" <porterboy76@yahoo.com> wrote:>> If you consider the analog signal bandlimited to B, then isn't it >> true at least in theory that if you analog sample the two >> inputs each at least 2*B you could then multiplex the digital >> portion? As long as the analog sampler meets Nyquist, you wouldn't >> lose anything, i.e., you could use those same digital algorithms >> once you get the signals into the digital domain. Am I missing >> something? > >Well, what about synchronisation? If you have a different AFE on each >of antennae, you have N different PLLs, N different ADCs and N >different AGCs. Digitally you are then trying to combine N different >signals which have different sampling phase/frequency offsets, >different carrier phase/frequency offsets, different symbol timing and >different nominal gains. In paticular there may be large differences in >the group delays, which means you would need digital buffering. Is this >not exteremely suboptimal (or at least very difficult to compensate >for). > >In fact, I guess I am making a lot of assumptions which may just be >untrue, since in general I havent a clue what happens before the signal >gets digitized... especially in MIMO. I'm a definite noob when it comes >to AFE stuff!As long as each channel is adequately samples I don't see a problem. One of the main ideas with MIMO is that the antennas are separated adequately so that the channel response seen at each antenna is largely independent of what is seen at the other antennas. It's expected that there will be phase/gain/etc. differences at each antenna, and the same phase difference that may happen due to the different sampling times could be achieved by moving the antenna, too, so I'd think it'd be a wash. Channel estimation for each antenna has to be done, anyway, so I don't think it'd be a problem. Good question, though... :) Eric Jacobsen Minister of Algorithms, Intel Corp. My opinions may not be Intel's opinions. http://www.ericjacobsen.org
Reply by ●November 10, 20062006-11-10
eric,
for how many years did you work on MIMO. tell frankly
particlereddy
Eric Jacobsen wrote:
> On 10 Nov 2006 01:03:34 -0800, "porterboy" <porterboy76@yahoo.com>
> wrote:
>
> >> If you consider the analog signal bandlimited to B, then isn't it
> >> true at least in theory that if you analog sample the two
> >> inputs each at least 2*B you could then multiplex the digital
> >> portion? As long as the analog sampler meets Nyquist, you wouldn't
> >> lose anything, i.e., you could use those same digital algorithms
> >> once you get the signals into the digital domain. Am I missing
> >> something?
> >
> >Well, what about synchronisation? If you have a different AFE on each
> >of antennae, you have N different PLLs, N different ADCs and N
> >different AGCs. Digitally you are then trying to combine N different
> >signals which have different sampling phase/frequency offsets,
> >different carrier phase/frequency offsets, different symbol timing and
> >different nominal gains. In paticular there may be large differences in
> >the group delays, which means you would need digital buffering. Is this
> >not exteremely suboptimal (or at least very difficult to compensate
> >for).
> >
> >In fact, I guess I am making a lot of assumptions which may just be
> >untrue, since in general I havent a clue what happens before the signal
> >gets digitized... especially in MIMO. I'm a definite noob when it comes
> >to AFE stuff!
>
> As long as each channel is adequately samples I don't see a problem.
>
> One of the main ideas with MIMO is that the antennas are separated
> adequately so that the channel response seen at each antenna is
> largely independent of what is seen at the other antennas. It's
> expected that there will be phase/gain/etc. differences at each
> antenna, and the same phase difference that may happen due to the
> different sampling times could be achieved by moving the antenna, too,
> so I'd think it'd be a wash.
>
> Channel estimation for each antenna has to be done, anyway, so I don't
> think it'd be a problem.
>
> Good question, though... :)
>
> Eric Jacobsen
> Minister of Algorithms, Intel Corp.
> My opinions may not be Intel's opinions.
> http://www.ericjacobsen.org
Reply by ●November 10, 20062006-11-10
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Eric Jacobsen wrote:> On 9 Nov 2006 05:27:23 -0800, porterboy76@yahoo.com wrote: > >>Does a MIMO receiver have a whole analog front end for each antenna? >>Or can you use one PLL, ADC and Filter, and switch between the antennae? >Well, our product descriptions certainly uses the phrases "Dual receiver antenna inputs" and "Maximum Ratio Combining [MRC] diversity to eliminate black spots and fading." make of that what you will :-) http://www.csr.com/products/unifirange.htm> Switching between antennas only achieves switch diversity, which is > good but not nearly as good as most MIMO algorithms. Most people > don't really call a switched system MIMO, just switched diversity. > e.g., 802.11b systems have had switch diversity in the receiver since > day one, but nobody calls them MIMO.Indeed.> > For the most part, yes, the receivers are separate. This is because > most of the effective combining/decoding algorithms are digital, so > each stream must be digitized separately. > > It would probably be possible to build something like an MRC receiver > that is analog so that the combining is done before the ADC, but that > would preclude other algorithms from being used.Well, combining signals from mutiple antennas in analogue (with different phase shifts in front of each contribution) is something i've seen used for beamforming, where the antennas are assumed to receive correlated signals. If the two channels are uncorrelated, however, what could you really hope to achieve reliably, other than 3 dB improvement in SNR in a noise-only environment? Best Regards Jens - -- Key ID 0x09723C12, jensting@tingleff.org Analogue filtering / 5GHz RLAN / Mandriva Linux / odds and ends http://www.tingleff.org/jensting/ +44 1223 211 585 "You didn't slay the dragon?" "It's on my TODO list!!" 'Shrek' -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2.2 (GNU/Linux) iD8DBQFFVN7zimJs3AlyPBIRAjhfAJ9HyNiG/RSzYHfUHG05sUKld7kMgwCg8qhH mSLUzHKGwlwCQxclaWlLzNs= =haXG -----END PGP SIGNATURE-----
Reply by ●November 10, 20062006-11-10
On Fri, 10 Nov 2006 20:19:57 +0000, Jens Tingleff <jensting@tingleff.org> wrote:>Eric Jacobsen wrote: >> >> For the most part, yes, the receivers are separate. This is because >> most of the effective combining/decoding algorithms are digital, so >> each stream must be digitized separately. >> >> It would probably be possible to build something like an MRC receiver >> that is analog so that the combining is done before the ADC, but that >> would preclude other algorithms from being used. > >Well, combining signals from mutiple antennas in analogue (with different >phase shifts in front of each contribution) is something i've seen used for >beamforming, where the antennas are assumed to receive correlated signals. >If the two channels are uncorrelated, however, what could you really hope >to achieve reliably, other than 3 dB improvement in SNR in a noise-only >environment?Not much, I'd think. Eric Jacobsen Minister of Algorithms, Intel Corp. My opinions may not be Intel's opinions. http://www.ericjacobsen.org
