60 GHz Impairments Modeling Month Year doc.: IEEE 802.11-07/xxxxr0 November 2009 60 GHz Impairments Modeling Date: 2009-11-19 Authors: Name Affiliations Address Phone Email Vinko Erceg Broadcom San Diego 858 521-5885 verceg@broadcom.com Murat Messe Alireza Tarighat Michael Boers Jason Trachewsky Changsoon Choi IHP choi@ihp-microelectronics.com Vinko Erceg, Broadcom Vinko Erceg, Broadcom

November 2009 Revision History Rev1: Phase Noise and CMOS PA AM-AM model parameters were updated Vinko Erceg, Broadcom

Outline PA non-linearities (distortion) modeling November 2009 Outline PA non-linearities (distortion) modeling PA Output Backoff (OBO) Phase noise modeling Carrier frequency offset and symbol clock modeling I/Q Imbalance modeling Vinko Erceg, Broadcom

PA Non-Linearities Modeling (1) November 2009 PA Non-Linearities Modeling (1) Input signal x(t) to an amplifier produces output signal y(t): Vinko Erceg, Broadcom

PA Non-Linearities Modeling (2) November 2009 PA Non-Linearities Modeling (2) Popular approaches to model G and Y are: Saleh Model [1] Both amplitude and phase distortion are modeled This model was originally developed for the Traveling Wave Tube Amplifiers (TWTA) For some parameter settings output power decreases as input power increases May be used for some Solid State Power Amplifier (SSPA) applications Rapp model [2] Originally developed to model only amplitude distortion Suitable for SSPA modeling Modified Rapp Model [3,5] Phase distortion modeling was added Ghorbani model [4] Both amplitude and phase distortion is modeled Vinko Erceg, Broadcom

November 2009 Ghorbani Model Both amplitude and phase distortions are modeled by 4 parameters: Vinko Erceg, Broadcom

November 2009 Rapp AM-AM Model Amplitude distortion (AM-AM) in Rapp model is modeled as: Vinko Erceg, Broadcom

Modified Rapp AM-PM Model November 2009 Modified Rapp AM-PM Model See reference [3] for modified Rapp model that includes also phase distortion modeling Phase distortion (AM-PM) may be also modeled as: The above equation is used for our modeling purposes Vinko Erceg, Broadcom

802.11n PA Distortion Model [7] Example November 2009 802.11n PA Distortion Model [7] Example IM1 PA non-linearity Simulation should be run at an oversampling rate of at least 4x. Use RAPP power amplifier model as specified in document 00/294 with p = 3. Calculate backoff as the output power backoff from full saturation: PA Backoff = ­10 log10(Average TX Power/Psat). Total TX power shall be limited to no more than 17 dBm. Disclose: (a) EIRP and how it was calculated, (b) PA Backoff, and (c) Psat per PA. Note: the intent of this IM is to allow different proposals to choose different output power operating points. Note: the value Psat = 25dBm is recommended. Note: AM-PM is not modeled Vinko Erceg, Broadcom

GaAs PA Model (1) GaAs PA Model November 2009 GaAs PA Model (1) GaAs PA Model In [5], a 802.15.3c PA distortion model was proposed based on the GaAs pHEMT 60GHz HPA measurements from NEC The NEC GaAs PA characteristics seem to have similar trend to other published/measured amplifier characteristics in this class Characteristic AM-AM and AM-PM curves Modified Rapp or Ghorbani models may be used for fitting the AM-AM and AM-PM experimental data points We use modified Rapp model Least squares fitting function Voltage is rms Highest voltage AM-PM point was not included in the modeling (does not follow trend) Vinko Erceg, Broadcom

November 2009 GaAs PA Model (2) Vinko Erceg, Broadcom

November 2009 GaAs PA Model (3) Vinko Erceg, Broadcom

GaAs PA Model (4) Modified Rapp model parameters for NEC GaAs PA November 2009 GaAs PA Model (4) Modified Rapp model parameters for NEC GaAs PA AM-AM parameters g = 19 Asat = 1.4 s = 0.81 AM-PM parameters a = - 48000 b = 0.123 q1 = 3.8 q2 = 3.7 Vinko Erceg, Broadcom

November 2009 CMOS PA Model (1) We use the measured data from a 65 nm CMOS 60 GHz PA in reference [6] Modified Rapp or Ghorbani models may be used for fitting the AM-AM and AM-PM experimental data points We use modified Rapp model Least squares fitting function Voltage is rms AM-PM response was normalized so that the first point has Phase = 0o Vinko Erceg, Broadcom

November 2009 CMOS PA Model (2) Vinko Erceg, Broadcom

November 2009 CMOS PA Model (3) Vinko Erceg, Broadcom

CMOS PA Model (4) Modified Rapp parameters for CMOS PA November 2009 CMOS PA Model (4) Modified Rapp parameters for CMOS PA AM-AM parameters g = 4.65 Asat = 0.58 s = 0.81 AM-PM parameters a = 2560 b = 0.114 q1 = 2.4 q2 = 2.3 Vinko Erceg, Broadcom

PA Output Backoff (1) PA Output Backoff (OBO) may be defined as: November 2009 PA Output Backoff (1) PA Output Backoff (OBO) may be defined as: where P is either PA saturation point or 1 dB PA compression point OBO is related to: Meeting spectrum mask requirements Increasing modulation accuracy (reducing EVM) Reducing Adjacent Channel Interference (ACI) Vinko Erceg, Broadcom

November 2009 PA Output Backoff (2) OBO values for OFDM system reported in [9-11] relative to the 1 dB PA compression point are approximately 6 dB for 64 QAM modulation with R = ¾ coding OBO value for OFDM system reported in [10] relative to the PA saturation point is approximately 9 dB for 64 QAM modulation with R = ¾ coding OBO values for OFDM system reported in [11] relative to the 1 dB PA compression point are approximately 4.5 dB for 16 QAM modulation with R = ¾ coding (performance degradation of 1.5 dB) Theoretical OBO value for Single Carrier (SC) GMSK modulation is 0 dB OBOSC_GMSK = 0.5 dB may be used Vinko Erceg, Broadcom

Modulation Accuracy (dB) November 2009 PA Output Backoff (3) OBO Requirement Spectrum Mask Requirements OBO (dB) Mod. Accuracy Requirements Modulation Accuracy (dB) EVM Vinko Erceg, Broadcom

November 2009 Phase Noise Model (1) Phase noise may be reasonably modeled by a two pole – one zero model We propose the following parameters of the model: PSD(0) = -90 dBc/Hz Pole frequency fp = 1 MHz Zero frequency fz = 100 MHz PSD(infinity) = -130 dBc/Hz Vinko Erceg, Broadcom

November 2009 Phase Noise Model (2) Vinko Erceg, Broadcom

Frequency Offset/Symbol Clock Accuracy November 2009 Frequency Offset/Symbol Clock Accuracy Symbol clock frequency tolerance in most systems is specified at +/- 20 ppm Reasonable cost/performance tradeoff Frequency offset of –13.675 ppm at the receiver, relative to the transmitter may be used [7] The symbol clock of the same relative offset as the carrier frequency offset may be used Vinko Erceg, Broadcom

I/Q Imbalance Modeling (1) November 2009 I/Q Imbalance Modeling (1) Following model may be used for I/Q imbalance modeling [8]: where y(t) is the ideal complex transmit signal, yd(t) is the distorted complex signal, and distortion coefficients are given as where θ and α are phase and gain imbalances, respectively Vinko Erceg, Broadcom

I/Q Imbalance Modeling (2) November 2009 I/Q Imbalance Modeling (2) Tx EVM (-dB) Vinko Erceg, Broadcom

I/Q Imbalance Modeling (3) November 2009 I/Q Imbalance Modeling (3) We propose that including I/Q imbalance in the simulations be optional Slides presented here regarding I/Q imbalance may serve as a reference Vinko Erceg, Broadcom

November 2009 Conclusion We propose the following impairments/parameters to be included in the simulations: PA distortion OBO Phase noise Frequency/Symbol Clock offset We propose that inclusion of the I/Q imbalance impairment is optional Vinko Erceg, Broadcom

November 2009 References [1] A.A.M. Saleh, "Frequency-independent and frequency-dependent nonlinear models of TWT amplifiers," IEEE Trans. Communications, vol. COM-29, November 1981, pp.1715-1720. [2] C. Rapp, "Effects of HPA-Nonlinearity on a 4-DPSK/OFDM-Signal for a Digital Sound Broadcasting System", in Proceedings of the Second European Conference on Satellite Communications, Liege, Belgium, Oct. 22-24, 1991, pp. 179-184. [3] M. Honkanen and Sven-Gustav Haggman, “New Aspects on Nonlinear Power Amplifier Modeling in Radio Communication System Simulations”, Proc. IEEE Int. Symp. On Personal, Indoor, and Mobile Comm, PIMRC ’97, Helsinki, Finland, Sep.1-4, 1997, pp. 844-848. [4] A. Ghorbani, and M. Sheikhan, “The effect of Solid State Power Amplifiers (SSPAs) Nonlinearities on MPSK and M-QAM Signal Transmission”, Sixth Int'l Conference on Digital Processing of Signals in Comm., 1991, pp. 193-197. [5] IEEE Document 15-06-0477-01-003c-rf-impairment-models-60ghz-band-sysphy-simulation.pdf. [6] Mikko Varonen, et. al. “Millimeter-Wave Amplifiers in 65-nm CMOS”. ESSCIRC 2007. 11-13 Sep. 2007. pp. 280-283. [7] IEEE Document 11-03-0814-31-000n-comparison-criteria.doc. [8] Alireza Tarighat, and Ali H. Sayed, “Joint Compensation of Transmitter and Receiver Impairments in OFDM Systems,” IEEE Transactions on Wireless Communications, VOL. 6, NO. 1, January 2007, pp. 240-247. Vinko Erceg, Broadcom

November 2009 References [9] Yongwang Ding and Ramesh Harjani, “A High-Efficiency CMOS +22-dBm Linear Power Amplifier,” IEEE Journal of Solid-State Circuits, VOL. 40, NO. 9, September 2005, pp. 1895-1900. [10] Mostafa Elmala, Jeyanandh Paramesh, and Krishnamurthy Soumyanath, “A 90-nm CMOS Doherty Power Amplifier With Minimum AM-PM Distortion,” IEEE Journal of Solid-State Circuits, VOL. 41, NO. 6, June 2006, pp. 1323-1332. [11] Mathias Pauli, Udo Wachsmann, Magnus Sundelin, and Peter Schramm, “Transmitter Impairments in OFDM-Based Wireless LAN,” Vehicular Technology Conference, 53rd VTC 2001 Spring, VOL 1,  6-9 May 2001, pp. 692 – 696. Vinko Erceg, Broadcom