Responses to “No” Vote Comments on Calibration

Responses to “No” Vote Comments on Calibration
May 2005 doc.: IEEE /0414r1 May 2005 Responses to “No” Vote Comments on Calibration Date: Authors: Notice: This document has been prepared to assist IEEE It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein. Release: The contributor grants a free, irrevocable license to the IEEE to incorporate material contained in this contribution, and any modifications thereof, in the creation of an IEEE Standards publication; to copyright in the IEEE’s name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEE’s sole discretion to permit others to reproduce in whole or in part the resulting IEEE Standards publication. The contributor also acknowledges and accepts that this contribution may be made public by IEEE Patent Policy and Procedures: The contributor is familiar with the IEEE 802 Patent Policy and Procedures < ieee802.org/guides/bylaws/sb-bylaws.pdf>, including the statement "IEEE standards may include the known use of patent(s), including patent applications, provided the IEEE receives assurance from the patent holder or applicant with respect to patents essential for compliance with both mandatory and optional portions of the standard." Early disclosure to the Working Group of patent information that might be relevant to the standard is essential to reduce the possibility for delays in the development process and increase the likelihood that the draft publication will be approved for publication. Please notify the Chair as early as possible, in written or electronic form, if patented technology (or technology under patent application) might be incorporated into a draft standard being developed within the IEEE Working Group. If you have questions, contact the IEEE Patent Committee Administrator at John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 Abstract This presentation addresses issues raised in “reasons and cures” comments addressing issues related to the calibration in theTGn Sync proposal John Ketchum, Qualcomm John Ketchum, Qualcomm

Comments on Calibration (1)
May 2005 doc.: IEEE /0414r1 May 2005 Comments on Calibration (1) John Ketchum, Qualcomm John Ketchum, Qualcomm

Summary of Calibration Comments
May 2005 doc.: IEEE /0414r1 May 2005 Summary of Calibration Comments Calibration should be optional Required frequency of calibration unclear or unknown Calibration will not work with independent AGC per Rx chain Gain and phase changes due to AGC action and drifts over temperature will create problems with calibration Calibration can only enforce reciprocity if the number of Tx and Rx chains at a STA is the same Calibration errors and resulting failure of reciprocity will remove any advantage of beamforming Calibration has not been proven to work with direct conversion receivers Reciprocity won’t hold with antenna coupling Reciprocity won’t hold with CCI Calibration interoperability will be a problem Need to calibrate will limit choices, overly constrain, and/or inhibit innovation in receiver design Calibration adds too much complexity Specify required accuracy and dimensionality of calibration Specify timing and control of calibration process Show that practical rf considerations wrt calibration will not overly constrain rf implementations John Ketchum, Qualcomm John Ketchum, Qualcomm

Responses to Comments on Calibration(1)
May 2005 doc.: IEEE /0414r1 May 2005 Responses to Comments on Calibration(1) Calibration should be optional TGn Sync has made calibration completely optional Uncalibrated client STAs can receive beamformed HT-Data with no performance degradation relative to calibrated client STAs. However, uncalibrated clients will not be able to effectively do transmit beamforming Required frequency of calibration unclear or unknown Extensive experience with discrete RFs shows that calibration is typically very stable over time intervals of at least hours Integrated RFs are expected to be better matched and easier to calibrate Typical receiver processing such as MMSE will clean up residual mis-matches due to calibration errors and other errors Beamforming performance degrades gracefully as calibration accuracy degrades. John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 Responses to Comments on Calibration(2) Calibration will not work with independent AGC per Rx chain Agree that independent AGCs are probably not a good idea with calibrated beamforming. STAs that do not support Tx beamforming and have independent AGCs can participate in calibration exchange and provide channel estimate to second STA that will allow second STA to become calibrated. Calibration can only enforce reciprocity if the number of Tx and Rx chains at a STA is the same It is true that you can only calibrate antennas that can be used for both Tx and Rx. Calibration is undefined for antennas that are Rx only (or Tx only) This is not a limitation, however, since STAs with more Rx than Tx antennas can still use the extra antennas to improve Rx performance, even if the corresponding STA is only steering to the calibrated antennas. John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 Responses to Comments on Calibration(3) Calibration errors and resulting failure of reciprocity will remove any advantage of beamforming Calibration performed under moderate SNRs still results in significant beamforming gains relative other approaches Mismatches due to calibration errors are easily corrected via receiver processing such as MMSE with minimal performance loss Calibration has not been proven to work with direct conversion receivers We don’t see any reason why calibration should be any more of a problem with well-designed ZIF receivers than with single- or double-conversion receivers. Constraints on good ZIF design for any MIMO approach (low DC offset and low I/Q imbalance) will result in high quality calibration. Reciprocity won’t hold with antenna coupling Antenna coupling doesn’t break resiprocity John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 Responses to Comments on Calibration(4) Reciprocity won’t hold with CCI CCI does not effect reciprocity Differing noise/interference levels at the two ends of a link only effect MCS choice, not optimum steering In cases of directional/spatially non-white interference, performance may benefit from interference cancellation/suppression measures John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 Responses to Comments on Calibration(5) Calibration interoperability will be a problem There is no reason for calibration to be an interoperability problem. To get accurate calibration, STAs only need to ensure channel estimates that have equivalent quality to the channel estimates that are used for Rx equalization. This is not a burden Need to calibrate will limit rf design choices, overly constrain rf implementations, and/or inhibit innovation in receiver design Calibration and beamforming undoubtedly impose constraints on receiver design, as does any other design choice. However these constraints are no more onerous or difficult than for other MIMO approaches The idea that this approach will inhibit innovation is in direct contradiction to all of the previous comments, the point of which is that it will be difficult to make this work. Difficult engineering problems engender innovation John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 Responses to Comments on Calibration(6) Calibration adds too much complexity Channel estimate required for calibration is the same channel estimate needed for Rx equalization/demodulation Calibration does not require real-time processing other than is normally required for a message exchange. Computations are not time critical. Transmission of resulting channel estimates is not time critical. Channel estimates are sent using existing management action frames, which are defined in both WWiSE and TGn Sync proposals MIMO Channel Response frame (WWiSE) Calibration Measurement Action frame (TGn Sync) John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 Responses to Comments on Calibration(7) Gain and phase changes due to AGC action and drifts over temperature will create problems with calibration Beamforming and calibration are optional For receive-only client (no Tx beamforming capability), lack of calibration does not have negative impact on performance Calibration is a low-overhead, background process. Need to recalibrate will not have a negative impact on t’put Need to account for calibration issues when designing RF for Tx-beamforming-capable devices. Specify required accuracy and dimensionality of calibration See simulation results to follow John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 Responses to Comments on Calibration(8) Specify timing and control of calibration process See subsequent material Show that practical rf considerations wrt calibration will not overly constrain rf implementations Calibration has been made optional Details in the following material support the practicality of calibration John Ketchum, Qualcomm John Ketchum, Qualcomm

Background on Calibration
May 2005 doc.: IEEE /0414r1 May 2005 Background on Calibration John Ketchum, Qualcomm John Ketchum, Qualcomm

Some Notation May 2005 May 2005 doc.: IEEE 802.11-05/0414r1
John Ketchum, Qualcomm John Ketchum, Qualcomm

Reciprocity May 2005 May 2005 doc.: IEEE 802.11-05/0414r1

Tx-side and Rx-side Calibration Correction
May 2005 doc.: IEEE /0414r1 May 2005 Tx-side and Rx-side Calibration Correction John Ketchum, Qualcomm John Ketchum, Qualcomm

Transmit-Side Correction
May 2005 doc.: IEEE /0414r1 May 2005 Transmit-Side Correction John Ketchum, Qualcomm John Ketchum, Qualcomm

Receive-side Correction
May 2005 doc.: IEEE /0414r1 May 2005 Receive-side Correction John Ketchum, Qualcomm John Ketchum, Qualcomm

Asymmetric Calibration
May 2005 doc.: IEEE /0414r1 May 2005 Asymmetric Calibration STA A does transmit correction STA B does receive correction John Ketchum, Qualcomm John Ketchum, Qualcomm

Asymmetric Calibration
May 2005 doc.: IEEE /0414r1 May 2005 Asymmetric Calibration John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 Uncalibrated Client STA does transmit correction, STA B does no correction John Ketchum, Qualcomm John Ketchum, Qualcomm

Single Antenna Client May 2005 May 2005 doc.: IEEE 802.11-05/0414r1

Spatial MMSE Receiver Basics
May 2005 doc.: IEEE /0414r1 May 2005 Spatial MMSE Receiver Basics John Ketchum, Qualcomm John Ketchum, Qualcomm

Beamforming to Uncalibrated Client
May 2005 doc.: IEEE /0414r1 May 2005 Beamforming to Uncalibrated Client John Ketchum, Qualcomm John Ketchum, Qualcomm

Beamforming to Calibrated Client
May 2005 doc.: IEEE /0414r1 May 2005 Beamforming to Calibrated Client John Ketchum, Qualcomm John Ketchum, Qualcomm

Calibration Exchange May 2005
doc.: IEEE /0414r1 May 2005 Calibration Exchange The Calibration Training frame exchange sequence operates as follows: STA A transmits a Calibration Training MPDU In response, STA B transmits a Calibration Training MPDU (Position field = 1) after a SIFS interval, which includes the HT-LTF that allows STA A to estimate the channel matrices. In response, STA A transmits a Calibration Training MPDU after a SIFS interval, which includes the HT-LTF that allows a receiving STA B to estimate the channel matrices. The remaining message exchange in the calibration procedure is not time critical. When the channel measurements become available at STA B, STA B sends one or more Calibration Measurement action frames that contains the MIMO Channel Measurement Report. STA B sets the Calibration Complete bit to ‘1’ in case it is not capable of transmit beamforming. In this case STA A will not send Calibration Correction frame. In case that both STAs are capable of transmit beamforming, once STA A has the MIMO Channel Measurement Report from STA B, along with its local channel measurement measured from the Calibration Training frame transmitted by STA B, STA A can compute the correction vectors required for both itself and STA B. The Calibration Correction frame is transmitted to complete the calibration procedure by STA A. The message includes the MIMO Correction Vector for STA B. John Ketchum, Qualcomm John Ketchum, Qualcomm

Calibration Exchange May 2005 May 2005 doc.: IEEE 802.11-05/0414r1

Measured Calibration Errors
May 2005 doc.: IEEE /0414r1 May 2005 Measured Calibration Errors Calibration based on a single channel measurement in each directions Measured calibration error is a strong function of SNR Resulting calibration used in operating modem Residual errors are compensated by receive processing John Ketchum, Qualcomm John Ketchum, Qualcomm

Achieved Average Calibration Error
May 2005 doc.: IEEE /0414r1 May 2005 Achieved Average Calibration Error Shows average of repeated calibration procedures Demonstrates that calibration error is primarily limited by receiver noise John Ketchum, Qualcomm John Ketchum, Qualcomm

Simulation Results for Calibration
May 2005 doc.: IEEE /0414r1 May 2005 Simulation Results for Calibration Matlab simulations using TGn channel models of six calibration conditions Noiseless estimates of Tx correction matrices for both forward and reverse links Tx steering computed from noiseless channel estimates Show Rx SNR and capacity with MMSE at both AP and client STA Rx Noisy estimates of Tx correction matrices for both forward and reverse links SNR on channel estimates for Tx correction matrices is the same as Rx SNR for demod. Tx steering computed from noisy channel estimates Two other client conditions Tx correction only at AP, but AP knows Tx correction for client STA, and applies it to RL channel estimate before computing Tx steering vectors Tx correction only at AP, no correction known for client STA Show Rx SNR and capacity with MMSE at client STA Rx John Ketchum, Qualcomm John Ketchum, Qualcomm

Simulation Conditions
May 2005 doc.: IEEE /0414r1 May 2005 Simulation Conditions TGn Channel models B and E (4x4 only) 2x2, 4x2, 4x4 15 dB and 25 dB Rx snr Rf imbalance on both Tx and Rx Uniformly distributed gain variation in dB of +/- 3 dB Uniformly distributed phase variation over +/- π radians Imbalance constant over the band 11-sample moving average smoothing of the calibration correction matrix in the frequency domain John Ketchum, Qualcomm John Ketchum, Qualcomm

2×2 Channel Model B SNR CDF
May 2005 doc.: IEEE /0414r1 May 2005 2×2 Channel Model B SNR CDF Principal Eigenmode Second eigenmode John Ketchum, Qualcomm John Ketchum, Qualcomm

2×2 Channel Model B Capacity CDF
May 2005 doc.: IEEE /0414r1 May 2005 2×2 Channel Model B Capacity CDF John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 4×2 Channel Model B SNR CDF John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 4×4 Channel Model B SNR CDF Principal Eigenmode 4nd eigenmode 2nd eigenmode 3nd eigenmode John Ketchum, Qualcomm John Ketchum, Qualcomm

4×4 Channel Model E SNR CDF
May 2005 doc.: IEEE /0414r1 May 2005 4×4 Channel Model E SNR CDF John Ketchum, Qualcomm John Ketchum, Qualcomm

4×4 Channel Model E Capacity CDF
May 2005 doc.: IEEE /0414r1 May 2005 4×4 Channel Model E Capacity CDF John Ketchum, Qualcomm John Ketchum, Qualcomm

May 2005 doc.: IEEE /0414r1 May 2005 Conclusion We have addressed issues raised in the “reasons and cures” comments relating to calibration Calibration is an optional functionality that supports an optional mode of operation Calibration is unnecessary for Rx-only devices to reliably receive PPDUs transmitted using beamforming Requirements for receive-only beamforming at client: AP is calibrated Requires that there is a client in the network that can participate in calibration OR Requires that AP can self-calibrate Client can send channel-sounding packet Requires client long training dimensionality == Ntx Simulation results show that beamforming performance is not significantly degraded by errors in computation of calibration correction vectors. Details of calibration methodologies have been provided John Ketchum, Qualcomm John Ketchum, Qualcomm

Responses to “No” Vote Comments on Calibration

Similar presentations

Presentation on theme: "Responses to “No” Vote Comments on Calibration"— Presentation transcript:

Similar presentations

About project

Feedback

Log in

Auth with social network:

Responses to “No” Vote Comments on Calibration

Similar presentations

Presentation on theme: "Responses to “No” Vote Comments on Calibration"— Presentation transcript:

Similar presentations

About project

Feedback