octofade Channel Emulation

Transcription

1 octofade Channel Emulation January Berlin Road, Bolton, MA ๐

2 Outline 2 What is channel emulation and why is it critical for MIMO systems? Channel modeling standards octofade-wifi software octofade 3GPP RTL

3 Channel Quality Wireless Channel 3 Frequency and time variable wireless channel Multipath creates a sum of multiple versions of the TX signal at the RX Mobility of reflectors and wireless devices causes Doppler-based fading Multiple antenna techniques are used to optimize transmission in the presence of multipath and Doppler fading MIMO=multiple input multiple output

4 Multipath and Flat Fading 4 In a wireless channel the signal propagating from TX to RX experiences Flat fading Multipath/Doppler fading Multipath reflections occur in clusters. +10 db 0 db Multipath fading component -15 db flat fading component Time

5 Wireless Channel Multipath cluster model 5 Composite angular spread Per path angular spread Composite angular spread Multipath and Doppler fading in the channel

6 Validating Radio DSP 6 A variety of channel conditions and complex multiple-antenna algorithms for adapting to these conditions make a channel emulator necessary for developing and testing radio DSP TX Channel Emulator Controlled programmable channel conditions Multipath Doppler Noise RX

7 RF Front End Development and Test of MIMO 7 Development and test of MIMO systems requires a channel emulator to emulate multipath and Doppler fading in a variety of wireless channels. Adaptive multiple antenna techniques, including TX and RX diversity, spatial multiplexing and beamforming involve sophisticated open and closed loop algorithms that must be tested under a range of controlled (emulated) channel conditions. Traditional channel emulators connect to DUTs conductively without antennas. Antennas and antenna arrays are part of the channel models. Digital IQ (CPRI/OBSAI) Channel Emulator DSP RF or digital IQ coupling to DUTs RF Front End Multiple RF modules can connect for scalability

8 SISO Channel Emulator Tapped Delay Line (TDL), H Input TX h 1 x Complex time-variable coefficients, h Delay 2 Delay 3 h 2 x + Output RX h K Delay K x where K is the number of taps in the TDL

9 4x4 MIMO Channel Emulator Logic 9 H 11 H 21 H 31 + M1 H 41 H 12 H 22 H 32 + M2 H 42 H 13 H 23 H 33 + M3 H 43 N1 N2 N3 N4 H 14 H 24 H 34 H 44 + M4

10 [h 1,1 1, h 1,1 2,, h 1,1 T ] [h 2,1 1, h 2,1 2,, h 2,1 T ] [h 1,2 1, h 1,2 2,, h 1,2 T ] H 1,1 H 2,1 + M1 [h 2,2 1, h 2,2 2,, h 2,2 T ] H 1,2 + M2 Tapped Delay Line, H n,m [h 1 1,3, h 2 1,3,, h T 1,3 ] [h 1 2,3, h 2 2,3,, h T 2,3 ] H 2,2 H 1,3 + M3 MIMO Channel Emulation Logic 2 x 4 Example [h 1,4 1, h 1,4 2,, h 1,4 T ] H 2,3 [h 2,4 1, h 2,4 2,, h 2,4 T ] Complex tap coefficients N1 N2 H 1,4 H 2,4 + M4

11 Complex Tap Coefficient Generator Doppler spectrum 1,1 k..k Tap coefficient factors 1,1 h 1,1 k..k Doppler spectrum 1,2 k..k Tap coefficient factors 1,2 h 1,2 k..k Doppler spectrum n,m k..k Spatial correlation matrix Polarization matrix Tap coefficient factors n,m h n,m k..k Doppler spectrum N,M k..k Tap coefficient factors N,M h N,M k..k where k is the tap number, K is the maximum number of taps and h is the time-variable coefficient

12 Doppler Spectrum Block FIR or IIR AWGN Generator Doppler filter Classical Bell Static Flat Rounded Gaussian Constant phase Butterworth Pure Doppler Specified for most 3GPP channel models; Classical = Jakes; Classical 3 db and 6 db Specified for n channel models; variations of this filter include Bell-spike, which is used by model F to model a 40 km/hour spike in the spectrum Models LOS on first tap; used in custom channel modeling Can also be implemented using RF attenuators via an identity matrix (i.e. connecting inputs to outputs through attenuators) Variations of this filter include Rounded 12 db Rician LOS component only, no Reilly Doppler filter

13 Notes on Doppler Filter Implementation AWGN sources are connected to Doppler filters that provide the desired spectral shape of the fading. The Doppler filter is IIR in the octofade implementation. For n models, the filter is Bell-shaped for models A through F and Bell-Spike for model F. The Bell spectrum models fading due to walking-speed motion in the environment at an average speed of 1.2 km/hr. The spike in the Bell-Spike spectrum adds the effect of a vehicle moving at an average speed of 40 km/hr. For 3GPP models, the Doppler spectrum is Classical.

14 Tap Coefficient Factors x + + x h n,m t K LOS component K 1 + K ejθ n, m NLOS component Fluorescent light P k PDP weighting Interpolation to TDL clock rate Tap coefficient factors n,m where t is the tap number, h is the time-variable complex coefficient, K is Rician K-factor

15 Notes on Tap Coefficient Factors The parameter K (Rician K-factor) determines the relative strength of the LOS and NLOS components and is set based on the chosen model. The K term models the LOS component. The 1 term models the NLOS 1+K 1+K component. The LOS component can only be present on the 1 st tap. If the distance between the transmitter and the receiver is greater than the distance to 1st reflector (typically a wall in the indoor environment ) then LOS component is not present. The presence of LOS is a configuration parameter that can be enabled or disabled. The first tap s LOS component isn t Doppler filtered. Thus, if LOS is present, the power spectrum on the first tap deviates from the Bell spectrum since it includes both the LOS and the NLOS components. If LOS is present, the PDF and CDF of the 1 st tap are Rician. If LOS is not present the PDF and CDF on the 1 st tap are Rayleigh. On all other taps the PDF and CDF are always Rayleigh. P t represents the Power Delay Profile (PDP) weighting, summed over all the clusters that contribute power for the t th tap. It reflects how strong the total power is at tap t.

16 Receiver Transmitter n/ac Correlation Matrix The spatial correlation matrix is a function of the angular spread of each cluster, angle of arrival (AoA) and angle of departure (AoD) n models assume that RX and TX antenna systems are uniform linear arrays with equally spaced antenna elements. Spatial correlation is implemented using the Kronecker product of the transmit and receive correlation matrices, R tx and R rx, respectively. These matrices are comprised of correlation coefficient terms,ρ, that depend on the PAS, AoA, AoD, tap powers and distance D between antenna elements. Fox computes the real and imaginary parts, R XX (D) and R XY (D), respectively, for each ρ. This allows spatial correlation based on the complex field (i.e., using ρ =R XX (D)+jR XY (D)) or real power (i.e., using ρ =R XX2 (D) +R XY2 (D)). Antenna element spacing, D Angle of arrival (AoA), Angle of departure (AoD) Antenna spacing, D AoD AoA

17 802.11ac Correlation and Polarization 17 MU-MIMO modeled for AoD and AoA Polarization matrix added since ac devices are expected to have cross-polarized antennas AoA = angle of arrival AoD = angle of departure

18 4x4 Uni-directional Fader Block Diagram Digitized 18 Fader logic Quad Quad RF I/O IQ IQ 4 RF Quad Atten Quad circulator RF RF I/O 4 Quad circulator Quad Atten RF return path 4

19 4x4 Bi-directional Fader Block Diagram Digitized 19 Fader Logic Quad Quad RF I/O IQ IQ 4 RF Quad Atten RF RF I/O 4 Quad circulator Quad Atten Quad Fader Logic Quad Quad circulator

20 Outline 20 What is channel emulation and why is it critical for MIMO systems? Channel modeling standards octofade-wifi software octofade-3gpp RTL

21 3GPP and Channel Models 21 Parameter Model Name References and Notes 3GPP Models (RTL) LTE: EPA 5Hz; EVA 5Hz; EVA 70Hz; ETU 70Hz; ETU 300Hz; High speed train; MBSFN GSM: RAx; HTx; TUx; EQx; TIx 3G: PA3; PB3; VA30; VA120; High speed train; Birth-Death propagation; Moving propagation; MBSFN 3GPP TS V ( ) 3GPP TS V ( ) 3GPP TS V ( ) Annex C 3GPP TS V ( ) 3GPP TS V ( ) 3GPP TS V ( ) IEEE n/ac Models (software) Channel modelling building blocks (RTL) A, B, C, D, E, F IEEE /940r4 IEEE Tap: delay, Doppler, PDP weight Path: list of taps System: NxM, correlation matrix

22 Channel Emulation Requirements Summary 22 RF bandwidth (no channel aggregation) EVM (avg downfading is -40 db) n ac 80 MHz 160 MHz LTE ( Annex B) 40 MHz 80 MHz 160 MHz 20 MHz -28 dbm (64QAM) -32 dbm (256QAM) -32 dbm (256QAM) -22 dbm (8% 64QAM) TDL Taps Delay resolution 10 ns 5 ns 2.5 ns 10 ns

23 Outline 23 What is channel emulation and why is it critical for MIMO systems? Channel modeling standards octofade-wifi software octofade-3gpp RTL

24 octofade Software 24 Up to 4x4 MIMO configuration n/ac channel emulator 4 x 4 Impairments: AWGN, spurious, phase noise, IQ imbalance, frequency shift Sample rate conversion 4 streams 4 streams Sample rate conversion Input file of sampled IQ streams Output file of sampled IQ streams AWGN = average white Gaussian noise

25 Use octofade Software with Off-the-shelf Equipment 25 octofade software Off the shelf VSA Off the shelf VSG Running on a Linux or Windows PC DUT

26 802.11n Channel Models - Summary 26 Model [1] Distance to 1 st wall (avg) # taps Delay spread (rms) Max delay A* test model 1 0 ns 0 ns B Residential 5 m 9 15 ns 80 ns 2 C small office 5 m ns 200 ns 2 D typical office 10 m ns 390 ns 3 E large office 20 m ns 730 ns 4 F large space (indoor or outdoor) 30 m ns 1050 ns 6 * Model A is a flat fading model; no delay spread and no multipath # clusters The LOS component is not present if the distance between the transmitter and the receiver is greater than the distance to 1 st wall. The presence of LOS is a configuration parameter that can be enabled or disabled.

27 802.11ac Channel Models ac channel models are an extension of n models [2] System Bandwidth W Channel Sampling Rate Expansion Factor Channel Tap Spacing W 40 MHz 1 10 ns 40 MHz < W 80 MHz 2 5 ns 80 MHz < W 160 MHz ns W > 160 MHz ns

28 octofade Software Architecture Model statistics (Matlab) CLI application + API API octofade channel modeling library

29 National Instruments LabVIEW Console 29 National Instruments LabVIEW application Graphical programming environment

30 Viewing Input and Output Streams 30 National Instruments TDMS file view TDMS = TDM streaming TDM = technical data management

31 Waveform Analysis 31 National Instruments WLAN Toolkit

32 Channel and Distortion Settings 32

33 802.11ac RTL DSP Requirements 33 Parameter Requirement Number of IQ input paths 1-8 IQ input data format 18-bit Number of IQ output paths 1-8 IQ output data format 18-bit Input/output sample rate Up to 400 MHz Channel bandwidth Up to 160 MHz Maximum number of total taps FPGA resources-dependent Number of TDL blocks Up to 64 Number of taps per TDL block FPGA resources-dependent Tap delay range FPGA resources-dependent Minimum tap delay resolution 2.5 ns Tap weight range 0 to -50 db Tap weight resolution 0.1dB Doppler shift 2 khz SNR setting -10 to +35 db, average +/- 0.1 db accuracy Noise filter bandwidth Up to 160 MHz Notes Unused inputs disabled where applicable Unused outputs set to zero where applicable Current capability: 100 MHz Current capability: 40 MHz N x M x taps_per_tdl Current capability: 30 usec Current capability: 10 ns To support high speed train Equal to preconfigured channel bandwidth Subject to the availability of FPGA resources, octoscope can customize any of these specifications.

34 RF Front End Considerations 34 Parameter Specification Notes MIMO configuration 8 x 8 To support emerging ac and LTE beamforming configurations Bidirectionality Important To support beamforming Bandwidth 160 MHz To support emerging ac Dynamic range (RF dynamic range, converter and DSP resolution) EVM Accommodate 52 db of output signal dynamic range with little distortion About 6 db higher than EVM required for RF signal Signal fluctuation: +9 db for PAPR +3 db for up-fade -40 db for down-fade For example, channel emulator s EVM should be at least 36 db over the entire dynamic range to minimize distortion of a 30 db EVM n signal

35 35 octofade-wi-fi Verification Examples

36 Doppler Spectrum n Model F Tap h Tap h Theoretical Simulated Frequency, Hz 2 Tap h 12 The Doppler spread is 3 Hz at 2.4 and 6 Hz at 5.25 GHz for environment speed of 1.2 km/hour Example of Doppler 10 0 spectrum plots for IEEE n model F 10 0 Environment velocity is 1.2 km/hour and is modeled on all taps for all models Tap 3 for model F includes automotive velocity spike at 40 km/hour Tap h 12

37 Doppler Spectrum n Model F, Tap Tap h FFT-based power estimation using entire long realization length 1024 periodograms with 50% overlap Doppler Spectra Frequency Frequency, Hz speed of light

38 Doppler Spectrum n Model E, Tap Tap h Fluorescent light effects at 120, 360, and 600 Hz harmonics of the power line frequency of 60 Hz Doppler Spectra Frequency Frequency, Hz

39 Cumulative Distribution Function (CDF) 0 Tx#1 - Rx# log 10 CDF -2-3 Tap 1 with LOS component IEEE n, Model F, Tx#2 CDF - for Rx#1 18 taps

40 Power Delay Profile (PDP) Model F Tx#1 - Rx#1 10 Power [db] Tap #, 1-18 Tap Power decreases with increasing tap delay Red points are for the normalized PDP under NLOS conditions. Blue points are simulated normalized PDP under LOS conditions. 10 Tx#2 - Rx#

41 Channel Impulse Response abs(h) time (s) Impulse response, IEEE n model F

42 Outline 42 What is channel emulation and why is it critical for MIMO systems? Channel modeling standards octofade-wifi software octofade-3gpp RTL

43 Generic models octofade System Architecture 43 octofade LabVIEW GUI LVlib wrapper n/ac Profile API SCME Profile API LTE Profile API WCDMA Profile API GSM Profile API Emulator API FPGA based DSP Host instrument hardware platform D/A and A/D Altera FPGA board RF Front End VST = vector signal transceiver

44 Generic models octofade Demo Configuration 44 octofade Demo VI PDP Verification VI DLL Library VIs Doppler Verification VI GSM Profile API WCDMA Profile API LTE Profile API Emulator API Emulator HAL (PCIe) Altera FPGA board

45 Demo Setup 45

46 Supported Models and Building Blocks 46 Parameter Model Name References and Notes 3GPP Models (RTL) LTE: EPA 5Hz; EVA 5Hz; EVA 70Hz; ETU 70Hz; ETU 300Hz; High speed train; MBSFN GSM: RAx; HTx; TUx; EQx; TIx 3G: PA3; PB3; VA30; VA120; High speed train; Birth-Death propagation; Moving propagation; MBSFN 3GPP TS V ( ) Annex B 3GPP TS V ( ) Annex B 3GPP TS V ( ) Annex C 3GPP TS V ( ) Annex B 3GPP TS V ( ) Annex B IEEE n/ac Models (software) A, B, C, D, E, F IEEE /940r4; IEEE Static Models (software and RTL) Channel modelling building blocks (RTL) Identity matrix Butler matrix Tap: delay, Doppler, PDP weight Path: list of taps System: NxM, correlation matrix Static bypass Static, minimum correlation

47 Parameter Digital IQ inputs (N) Digital IQ outputs (M) MIMO channels octofade-3gpp RTL DSP Specifications Specification Up to 8 (16-bit) Up to 4 (16-bit) Up to 2 independently configured MIMO channels 47 Maximum number of TDL paths Maximum number of taps per TDL Maximum number of taps Tap specifications 3GPP channel models Custom channel models Distortion AWGN Fading Control Interface 16 paths 18 taps per TDL 144 taps total 0 to -40dB; 0.1dB resolution 10ns delay resolution 30us max delay range LTE: EPA 5Hz; EVA 5Hz;EVA 70Hz; ETU 70Hz; ETU 300Hz; MBSFN GSM: RAx; HTx; TUx; EQx; TIx UMTS: PA3; PB3; VA30; VA120; MBSFN Moving propagation, birth-death, high-speed train Arbitrary fading profiles can be implemented in software and loaded into the FPGA via register interface. Better than -50dBc SNR: -10 to +25dB; 0.1dB resolution Bandwidth: Up to 20MHz, per 3GPP standard requirements Types: Rician/Rayleigh Spectrum: Classical Doppler Shift: Hz; 1Hz resolution 32-bit register memory map

48 octofade-module 48 octofade-module hardware is based on off-the-shelf Altera Stratix-IV FPGA board with octofade-3gpp RTL integrated into the on-board FPGA - Highly efficient implementation with only 1/3 rd of the FPGA resource used Formfactor IQ interface FPGA resources* Stratix IV EP4SGX530KH40C2 Board memory DDR3/SSRAM x36* PCI Express Interface PCIe DDR HSMC Ports A and B Clock rate: MHz Bit resolution: 16 bits I, 16 bits Q Logic ALUTs: 97,616 used; 424,960 available (23% used) memory bits: 7,760,493 MB used; Available 21,233,664 (37% used) 18-bit DSP Blocks: 296 used; Available 1,024 (29% used) 0 MB used; 512 MB available x4 PCI Express 4-lane interface to host PC for configuration and verification

49 FPGA Resource Estimate 49 Altera Stratix IV EP4Sx360 Xilinx Virtex 6 XC6VLX195T Logic 104K ALUTs, 62K Regs 73K ALMs 88K - 173K LUT % MEMORY x36 = 9kb blocks Kb blocks 34% 48 2K x 72= 148kb blocks Kb blocks 56% DSP 296 dual multiply/add blocks 592 DSP slices 93%

50 50 octofade-3gpp Logic Verification Examples

51 3GPP LTE PDP MIMO 4x4 Example 51 LTE ETU model Tx#1 - Rx#1 Tx#1 - Rx#2 Tx#1 - Rx#3 Tx#1 - Rx#4 Power [db] theo sim Tx#2 - Rx#1 Tx#2 - Rx#2 Tx#2 - Rx#3 Tx#2 - Rx#4 Power [db] Tx#3 - Rx#1 Tx#3 - Rx#2 Tx#3 - Rx#3 Tx#3 - Rx#4 Power [db] Tx#4 - Rx#1 Tx#4 - Rx#2 Tx#4 - Rx#3 Tx#4 - Rx#4 Power [db] Tap index Tap index Tap index Tap index Source: octofade Verification Report PDP = power delay profile

52 3GPP Doppler Spectrum Matlab Model h 1, sim 40 Spectrum theory Norm. Freq. Source: octofade Verification Report

53 3GPP Tap Coefficients CDF Example 53 0 Tx#1 - Rx# log 10 CDF Rayleigh log (h) [db] 10 Source: octofade Verification Report CDF cumulative distribution function

54 3GPP Moving Propagation Example Moving Propagation - Tap 2 tap index time (s) Source: octofade Verification Report

55 3GPP Birth-Death Example histogram of path number of occurrences tap index Source: octofade Verification Report

56 References 1. IEEE, /940r4: TGn Channel Models; May 10, IEEE, , TGac Channel Model Addendum Supporting Material, May IEEE, ad-channel-models-for-60-ghz-wlan-systems 4. Schumacher et al, "Description of a MATLAB implementation of the Indoor MIMO WLAN channel model proposed by the IEEE TGn Channel Model Special Committee", May Schumacher et al, "From antenna spacings to theoretical capacities - guidelines for simulating MIMO systems" 6. Schumacher reference software for implementing and verifying n models TS , Annex B, User Equipment (UE) radio transmission and reception (FDD), file: b00-AnnexB.doc 8. TS , Annex B, Evolved Universal Terrestrial Radio Access (E-UTRA); User Equipment (UE) radio transmission and reception, file: a50- AnnexB.doc 9. TS , Annex B, Evolved Universal Terrestrial Radio Access (E-UTRA); User Equipment (UE) conformance specification Radio transmission and reception Part 1: Conformance Testing, file: a00_s09-sAnnexB.doc 10. TS , Annex C, GSM/EDGE Radio Access Network; Radio transmission and reception, file: a30-AnnexC.doc 11. TS , Mobile Station (MS) conformance specification; Part 1: Conformance specification, file: _s00-s11.doc 12. 3GPP TR , "3rd Generation Partnership Project; technical specification group radio access networks; Spatial channel model for MIMO simulations 13. IST-WINNER II Deliverable v.1.2, WINNER II Channel Models, IST-WINNER2, Tech. Rep., 2008 ( GPP TS : User Equipment (UE) / Mobile Station (MS) Over The Air (OTA) Antenna Performance Conformance Testing 15. CTIA, Test Plan for Mobile Station Over the Air Performance - Method of Measurement for Radiated RF Power and Receiver Performance, Revision 3.1, January 2011

57 Thank you! 57 octofade web page Contact Fanny Mlinarsky Mobile: