Hi, I'm trying to implement a Wimax system using beamforming (8 antennas at the base station) with spatial multiplexing (two antennas at the MS), so that I would end up with two spatial "pipes". The base station would therefore use a four-antenna array to steer a beam to the first receive antenna, and another array to steer a beam to the second one. I am using Matlab. I already have written the spatial multiplexing part, but I still doesn't have a clue on how to do the beamforming. Could somebody please give me tips on how to implement the beamforming part? Any hints on good reference material will be appreciated.
Simulating beamforming with MIMO in WiMAX
Started by ●June 5, 2008
Reply by ●June 12, 20082008-06-12
On Jun 5, 6:56 pm, "citizen4" <jah...@gmail.com> wrote:> Hi, > > I'm trying to implement a Wimax system using beamforming (8 antennas at > the base station) with spatial multiplexing (two antennas at the MS), so > that I would end up with two spatial "pipes". The base station would > therefore use a four-antenna array to steer a beam to the first receive > antenna, and another array to steer a beam to the second one. I am using > Matlab. > > I already have written the spatial multiplexing part, but I still doesn't > have a clue on how to do the beamforming. Could somebody please give me > tips on how to implement the beamforming part? Any hints on good reference > material will be appreciated.Lets assume that you have the minimal 2 receive antennas required. So you have a 2 by 8 matrix. Do a SVD on that and you get a 2x2 U, 2x8 D and a 8x8 V^H. Your beamforming matrix is the first two columns of V. If have more receive antennas, you can still choose the first two columns of V to transmit on the largest two eigenmodes. You can refer to any papers that do SVD beamforming. -K
Reply by ●August 18, 20082008-08-18
>Lets assume that you have the minimal 2 receive antennas required. So >you have a 2 by 8 matrix. Do a SVD on that and you get a 2x2 U, 2x8 D >and a 8x8 V^H. Your beamforming matrix is the first two columns of V. >If have more receive antennas, you can still choose the first two >columns of V to transmit on the largest two eigenmodes. You can refer >to any papers that do SVD beamforming. > >-K >I understand this is called mathematical beamforming or eigenbeamforming. Comparing to other (direction of arrival) based beamforming methods, is this technique actually possible to be implemented in real life systems (such as WiMAX here)?
Reply by ●August 18, 20082008-08-18
On Mon, 18 Aug 2008 05:53:32 -0500, "citizen4" <jaheti@gmail.com> wrote:> >>Lets assume that you have the minimal 2 receive antennas required. So >>you have a 2 by 8 matrix. Do a SVD on that and you get a 2x2 U, 2x8 D >>and a 8x8 V^H. Your beamforming matrix is the first two columns of V. >>If have more receive antennas, you can still choose the first two >>columns of V to transmit on the largest two eigenmodes. You can refer >>to any papers that do SVD beamforming. >> >>-K >> > >I understand this is called mathematical beamforming or eigenbeamforming. > >Comparing to other (direction of arrival) based beamforming methods, is >this technique actually possible to be implemented in real life systems >(such as WiMAX here)?There's been a lot of controversy around just that. For cost-sensitive consumer devices (e.g., WiMAX user terminal, WiFi equipment) I think it's a tough case to make that the complexity is justified. That seemed to be the hallway scuttlebutt that I kept hearing among many 802.16 and 802.11 attendees when this stuff was being written into those specs. Others were enamored by simulations showing 10dB (!) of gain in some conditions and therefore thought it was a must-have. On the other hand people thought OFDM was too complex to implement for a long time, too. So if complexity costs and power consumption get manageable enough then it may well start appearing in common devices. The Turbo/LDPC codes are complex, too, and they give consistent gain regardless of the channel conditions. Nevertheless those are seldom implemented, either, due to the added complexity. Eric Jacobsen Minister of Algorithms Abineau Communications http://www.ericjacobsen.org Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php
Reply by ●August 25, 20082008-08-25
It seems to me that there isn't one definite way to apply beamforming (or AAS, as it's called in the standard) to a WiMAX system. According to what I've read so far, the only part of AAS that's in the standard is the part that defines how the channel information is fed back to the base station. Thus, it would be up to the system designer to come up with the algorithm to calculate the beamforming coefficients at the base station (and mobile station, too). Is this correct?
Reply by ●August 25, 20082008-08-25
On Mon, 25 Aug 2008 00:53:40 -0500, "citizen4" <wimaxeng@gmail.com> wrote:>It seems to me that there isn't one definite way to apply beamforming (or >AAS, as it's called in the standard) to a WiMAX system. According to what >I've read so far, the only part of AAS that's in the standard is the part >that defines how the channel information is fed back to the base station. >Thus, it would be up to the system designer to come up with the algorithm >to calculate the beamforming coefficients at the base station (and mobile >station, too). > >Is this correct?I'm not intimately familiar with the AAS part of the spec, but it sounds right just from the general philosophy of wireless standards. Generally all that is defined in the standard is the "air interface", or things that affect the transmitted waveform or control signalling. Implementation is up to the implementers, so implementation is generally not specified unless required for interoperation. In the case of AAS the algorithm/implementation doesn't need to be specified, only the feedback. So it sounds right to me. Eric Jacobsen Minister of Algorithms Abineau Communications http://www.ericjacobsen.org Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php
