Reply by Eric Jacobsen ●April 21, 20072007-04-21
On Fri, 20 Apr 2007 13:23:26 -0500, "philgo" <philgo@gmail.com> wrote:


>Let me try to rephrase the question.
>
>In traditional beam-forming, we are always talking about angles, angle of
>arrival, angle of departure, beam pattern pointing to a certain angle,
>etc.
>However, when we talk about the MRC, the concept of angle disappears and
>what we cared is only maximizing the SNR (or some other related metric). 
>
>Why we do not see/talk about the angles any more in MRC? I try to answer
>this question. Please share your comments and criticism.


I'll share my intuitive perspective and see whether it helps anything:

Beamforming requires coherence (i.e., phase alignment) between the
inputs to form and steer the beam as desired.

MRC only adjusts the inputs with a scale factor, so no beamforming is
taking place.   That's good because, as you've stated, each antenna
may be collecting energy from different reflected directions, so
forming a coherent beam may reduce the total amount of energy
collected and hurt rather than help.

Beamforming discriminates the direction in which energy is collected
by adjusting relative phases.   Some energy may be rejected in the
process.

MRC uses all of the energy collected by all antennas combined to
maximize SNR.

They're two very different techniques, but each have their purpose and
place.

I like the simplicity and elegance of MRC, though.   The fact that
it's more or less optimal in some conditions is just icing on the
cake.   ;)

Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org
Reply by Oli Charlesworth ●April 20, 20072007-04-20
philgo said the following on 20/04/2007 19:23:

> Let me try to rephrase the question.
> 
> In traditional beam-forming, we are always talking about angles, angle of
> arrival, angle of departure, beam pattern pointing to a certain angle,
> etc.
> However, when we talk about the MRC, the concept of angle disappears and
> what we cared is only maximizing the SNR (or some other related metric). 
> 
> Why we do not see/talk about the angles any more in MRC? I try to answer
> this question. Please share your comments and criticism.


A very close analogy is that of Wiener deconvolution 
(http://en.wikipedia.org/wiki/Wiener_deconvolution), where we're trying 
to invert an LTI system h(t).  Naively, we might think to do this in the 
frequency domain simply by multiplying by 1/H(f) (where H(f) <=> h(t)). 
  However, the optimal solution from an SNR point of view is the Wiener 
deconvolution.

Similarly, in antenna arrays, the optimal SNR solution is MRC (which is 
also a Wiener solution), not "naive" beamforming.

However, in both situations, when the noise is removed, the Wiener/MRC 
solution reduces to the "naive" solution.


-- 
Oli
Reply by Randy Yates ●April 20, 20072007-04-20
Phil,

It just so happens that I'm taking Information Theory this semester and we
just covered capacity of a MIMO system in fading channels. Your explanation
closely matches the professor's. He wouldn't call it "beamforming gain" for
the same reasons you stated.

--Randy


"philgo" <philgo@gmail.com> writes:


> Let me try to rephrase the question.
>
> In traditional beam-forming, we are always talking about angles, angle of
> arrival, angle of departure, beam pattern pointing to a certain angle,
> etc.
> However, when we talk about the MRC, the concept of angle disappears and
> what we cared is only maximizing the SNR (or some other related metric). 
>
> Why we do not see/talk about the angles any more in MRC? I try to answer
> this question. Please share your comments and criticism.
>
> In traditional beamforming, we often are in a line-of-sight scenario,
> e.g., in application of satellite, radar, etc. Remember we always talk
> about a incident plane wave in the beamforming textbook? In such cases,
> the channel is more like a Rician channel in mobile communication
> terminology. If we carry out MRC in such a scenario, the solution will be
> the same as what traditional beamforming will give us. And in such cases,
> given the simple LOS channel, the beamforming vector does reflect the
> actual geometry via the angle of arrival/departure. That's why we can talk
> about angles in traditional beamforming.
>
> In MRC, we often are in a non-line-of-sight scenario with fading not
> negligible. There are probably multi-path involved, with more than paths
> arriving in different directions. The channel is more like Rayleigh
> (Rician channel with small Rician factor K) channel in mobile
> communication terminology. MRC is still the optimal solution. However,
> given that multiple paths arriving from different directions, we can not
> follow up on the geometry any more. And that's why we no longer talk about
> angles any more in general MRC.
>
> On the other hand, I think we can still plot beam patterns in MRC case.
> The problem is that, the beam pattern may not reflect the actual geometry
> very well. By actual geometry, I mean actual position/direction of where
> the transmitter/receiver is. 
>
> I am not sure whether this is true or not. Please let me know your
> comments.
>
>>On Mar 23, 9:58 am, "Mike Yarwood" <mpyarw...@btopenworld.com> wrote:
>>> "dgse" <dgsei...@gmail.com> wrote in message
>>>
>>> news:1174630280.129018.82510@l77g2000hsb.googlegroups.com...> On Mar
> 23, 12:16 pm, "Rune Allnor" <all...@tele.ntnu.no> wrote:
>>> >> On 23 Mar, 03:09, "dgse" <dgsei...@gmail.com> wrote:
>>>
>>> >> > Q1:In receiver side, what is difference between maximum ratio
> combing
>>> >> > and adaptive beamforming?
>>> >> > Q2: In transmitter side, what is difference between maximum ratio
>>> >> > combing and adaptive beamforming?
>>>
>>> >> Homework?
>>>
>>> > No!
>>>
>>> I have never heard of maximum ratio combing. Do you have a reference?
>>>
>>> Best of luck - Mike
>>
>>A nice ap note from a quick Google:
>>http://www.freescale.com/files/dsp/doc/app_note/AN2251.pdf
>>This begins with examples of MRC concepts in the System Model section.
>>
>>An example:
>>http://www.iss.rwth-aachen.de/4_publikationen/res_pdf/Schmitt_ISSSTA_2004.pdf
>>
>>And a definition:
>>http://encyclopedia.thefreedictionary.com/Maximum+ratio+combining
>>
>>MRC seems to be about combining schemes for 'channels'. 'Channels'
>>might be, but need not be beams. MRC is an operation adapting on
>>signal quality to optimize BER, not form a spatial response pattern.
>>So. the answer to Q1 and Q2 is that beamformers form beams and MRC
>>doesn't care whether channels are beams, which is why the original
>>question seems ill-formed.
>>
>>A more useful response might be that there could be scenarios where
>>the MRC process produces a beamformer, but the MRC algorithm might not
>>know or care.
>>
>>Dale B. Dalrymple
>>http://dbdimages.com
>>
>>
>
>
>
>
>
> _____________________________________
> Do you know a company who employs DSP engineers?  
> Is it already listed at http://dsprelated.com/employers.php ?


-- 
%  Randy Yates                  % "Though you ride on the wheels of tomorrow,
%% Fuquay-Varina, NC            %  you still wander the fields of your
%%% 919-577-9882                %  sorrow."
%%%% <yates@ieee.org>           % '21st Century Man', *Time*, ELO
http://home.earthlink.net/~yatescr
Reply by philgo ●April 20, 20072007-04-20
Let me try to rephrase the question.

In traditional beam-forming, we are always talking about angles, angle of
arrival, angle of departure, beam pattern pointing to a certain angle,
etc.
However, when we talk about the MRC, the concept of angle disappears and
what we cared is only maximizing the SNR (or some other related metric). 

Why we do not see/talk about the angles any more in MRC? I try to answer
this question. Please share your comments and criticism.

In traditional beamforming, we often are in a line-of-sight scenario,
e.g., in application of satellite, radar, etc. Remember we always talk
about a incident plane wave in the beamforming textbook? In such cases,
the channel is more like a Rician channel in mobile communication
terminology. If we carry out MRC in such a scenario, the solution will be
the same as what traditional beamforming will give us. And in such cases,
given the simple LOS channel, the beamforming vector does reflect the
actual geometry via the angle of arrival/departure. That's why we can talk
about angles in traditional beamforming.

In MRC, we often are in a non-line-of-sight scenario with fading not
negligible. There are probably multi-path involved, with more than paths
arriving in different directions. The channel is more like Rayleigh
(Rician channel with small Rician factor K) channel in mobile
communication terminology. MRC is still the optimal solution. However,
given that multiple paths arriving from different directions, we can not
follow up on the geometry any more. And that's why we no longer talk about
angles any more in general MRC.

On the other hand, I think we can still plot beam patterns in MRC case.
The problem is that, the beam pattern may not reflect the actual geometry
very well. By actual geometry, I mean actual position/direction of where
the transmitter/receiver is. 

I am not sure whether this is true or not. Please let me know your
comments.


>On Mar 23, 9:58 am, "Mike Yarwood" <mpyarw...@btopenworld.com> wrote:
>> "dgse" <dgsei...@gmail.com> wrote in message
>>
>> news:1174630280.129018.82510@l77g2000hsb.googlegroups.com...> On Mar

23, 12:16 pm, "Rune Allnor" <all...@tele.ntnu.no> wrote:

>> >> On 23 Mar, 03:09, "dgse" <dgsei...@gmail.com> wrote:
>>
>> >> > Q1:In receiver side, what is difference between maximum ratio

combing

>> >> > and adaptive beamforming?
>> >> > Q2: In transmitter side, what is difference between maximum ratio
>> >> > combing and adaptive beamforming?
>>
>> >> Homework?
>>
>> > No!
>>
>> I have never heard of maximum ratio combing. Do you have a reference?
>>
>> Best of luck - Mike
>
>A nice ap note from a quick Google:
>http://www.freescale.com/files/dsp/doc/app_note/AN2251.pdf
>This begins with examples of MRC concepts in the System Model section.
>
>An example:
>http://www.iss.rwth-aachen.de/4_publikationen/res_pdf/Schmitt_ISSSTA_2004.pdf
>
>And a definition:
>http://encyclopedia.thefreedictionary.com/Maximum+ratio+combining
>
>MRC seems to be about combining schemes for 'channels'. 'Channels'
>might be, but need not be beams. MRC is an operation adapting on
>signal quality to optimize BER, not form a spatial response pattern.
>So. the answer to Q1 and Q2 is that beamformers form beams and MRC
>doesn't care whether channels are beams, which is why the original
>question seems ill-formed.
>
>A more useful response might be that there could be scenarios where
>the MRC process produces a beamformer, but the MRC algorithm might not
>know or care.
>
>Dale B. Dalrymple
>http://dbdimages.com
>
>






_____________________________________
Do you know a company who employs DSP engineers?  
Is it already listed at http://dsprelated.com/employers.php ?
Reply by dgse ●March 24, 20072007-03-24
On Mar 24, 11:54 am, "dbd" <d...@ieee.org> wrote:

> On Mar 23, 9:58 am, "Mike Yarwood" <mpyarw...@btopenworld.com> wrote:
>
>
>
> > "dgse" <dgsei...@gmail.com> wrote in message
>
> >news:1174630280.129018.82510@l77g2000hsb.googlegroups.com...> On Mar 23, 12:16 pm, "Rune Allnor" <all...@tele.ntnu.no> wrote:
> > >> On 23 Mar, 03:09, "dgse" <dgsei...@gmail.com> wrote:
>
> > >> > Q1:In receiver side, what is difference between maximum ratio combing
> > >> > and adaptive beamforming?
> > >> > Q2: In transmitter side, what is difference between maximum ratio
> > >> > combing and adaptive beamforming?
>
> > >> Homework?
>
> > > No!
>
> > I have never heard of maximum ratio combing. Do you have a reference?
>
> > Best of luck - Mike
>
> A nice ap note from a quick Google:http://www.freescale.com/files/dsp/doc/app_note/AN2251.pdf
> This begins with examples of MRC concepts in the System Model section.
>
> An example:http://www.iss.rwth-aachen.de/4_publikationen/res_pdf/Schmitt_ISSSTA_...
>
> And a definition:http://encyclopedia.thefreedictionary.com/Maximum+ratio+combining
>
> MRC seems to be about combining schemes for 'channels'. 'Channels'
> might be, but need not be beams. MRC is an operation adapting on
> signal quality to optimize BER, not form a spatial response pattern.
> So. the answer to Q1 and Q2 is that beamformers form beams and MRC
> doesn't care whether channels are beams, which is why the original
> question seems ill-formed.
>
> A more useful response might be that there could be scenarios where
> the MRC process produces a beamformer, but the MRC algorithm might not
> know or care.
>
> Dale B. Dalrymplehttp://dbdimages.com



Thanks dbd
Basically, what I am understand is for MRC the weight is real number
and for beamforming the weight is complex number.
Reply by dbd ●March 24, 20072007-03-24
On Mar 23, 9:58 am, "Mike Yarwood" <mpyarw...@btopenworld.com> wrote:

> "dgse" <dgsei...@gmail.com> wrote in message
>
> news:1174630280.129018.82510@l77g2000hsb.googlegroups.com...> On Mar 23, 12:16 pm, "Rune Allnor" <all...@tele.ntnu.no> wrote:
> >> On 23 Mar, 03:09, "dgse" <dgsei...@gmail.com> wrote:
>
> >> > Q1:In receiver side, what is difference between maximum ratio combing
> >> > and adaptive beamforming?
> >> > Q2: In transmitter side, what is difference between maximum ratio
> >> > combing and adaptive beamforming?
>
> >> Homework?
>
> > No!
>
> I have never heard of maximum ratio combing. Do you have a reference?
>
> Best of luck - Mike


A nice ap note from a quick Google:
http://www.freescale.com/files/dsp/doc/app_note/AN2251.pdf
This begins with examples of MRC concepts in the System Model section.

An example:
http://www.iss.rwth-aachen.de/4_publikationen/res_pdf/Schmitt_ISSSTA_2004.pdf

And a definition:
http://encyclopedia.thefreedictionary.com/Maximum+ratio+combining

MRC seems to be about combining schemes for 'channels'. 'Channels'
might be, but need not be beams. MRC is an operation adapting on
signal quality to optimize BER, not form a spatial response pattern.
So. the answer to Q1 and Q2 is that beamformers form beams and MRC
doesn't care whether channels are beams, which is why the original
question seems ill-formed.

A more useful response might be that there could be scenarios where
the MRC process produces a beamformer, but the MRC algorithm might not
know or care.

Dale B. Dalrymple
http://dbdimages.com
Reply by Mike Yarwood ●March 23, 20072007-03-23
"dgse" <dgseigig@gmail.com> wrote in message 
news:1174630280.129018.82510@l77g2000hsb.googlegroups.com...

> On Mar 23, 12:16 pm, "Rune Allnor" <all...@tele.ntnu.no> wrote:
>> On 23 Mar, 03:09, "dgse" <dgsei...@gmail.com> wrote:
>>
>> > Q1:In receiver side, what is difference between maximum ratio combing
>> > and adaptive beamforming?
>> > Q2: In transmitter side, what is difference between maximum ratio
>> > combing and adaptive beamforming?
>>
>> Homework?
>>
> No!
>

I have never heard of maximum ratio combing. Do you have a reference?

Best of luck - Mike
Reply by julius ●March 23, 20072007-03-23
On Mar 22, 9:09 pm, "dgse" <dgsei...@gmail.com> wrote:

> Q1:In receiver side, what is difference between maximum ratio combing
> and adaptive beamforming?
> Q2: In transmitter side, what is difference between maximum ratio
> combing and adaptive beamforming?



Realize that your question is not well-formed.
There are many flavors of "adaptive beamforming",
which one are you talking about?

Julius
Reply by dgse ●March 23, 20072007-03-23
No!

On Mar 23, 12:16 pm, "Rune Allnor" <all...@tele.ntnu.no> wrote:

> On 23 Mar, 03:09, "dgse" <dgsei...@gmail.com> wrote:
>
> > Q1:In receiver side, what is difference between maximum ratio combing
> > and adaptive beamforming?
> > Q2: In transmitter side, what is difference between maximum ratio
> > combing and adaptive beamforming?
>
> Homework?
>
> Rune
Reply by Rune Allnor ●March 23, 20072007-03-23
On 23 Mar, 03:09, "dgse" <dgsei...@gmail.com> wrote:

> Q1:In receiver side, what is difference between maximum ratio combing
> and adaptive beamforming?
> Q2: In transmitter side, what is difference between maximum ratio
> combing and adaptive beamforming?


Homework?

Rune