Channel Modelling g ETI 085

Lecture no:

Why directional channel models?

The spatial domain can be used to increase the spectral efficiency ffi i of f th the system t
Smart antennas MIMO systems

Directional channel models Channel sounding

Fredrik Tufvesson Department of Electrical and Information Technology Lund University, Sweden Fredrik.Tufvesson@eit.lth.se
2008-11-20 Fredrik Tufvesson - ETI 085 1

Need to know directional properties

How many significant reflection points? Which directions?

Model independent p on specific p antenna p pattern

2008-11-20

Fredrik Tufvesson - ETI 085

Double directional impulse response

TX position RX position
N r 1

Physical interpretation

number of multipath components for th f these positions iti

h t , r TX , r RX , , , h t , r TX , r RX , , ,
direction of arrival direction-of-arrival delay direction-of-departure

h t , r TX , r RX , , , |a |e j

2008-11-20

Fredrik Tufvesson - ETI 085

2008-11-20

Fredrik Tufvesson - ETI 085

Angular spread
E s , , , s , , , P s , , ,
double directional delay power spectrum

Directional models
The double directional delay power spectrum is sometimes factorized w r t DoD w.r.t. DoD, DoA and delay delay.

DDDPS, , APS BS APS MS PDP

Often in reality there are groups of scatterers with similar DoD and DoA clusters c c,BS c,MS c DDDPS , , P k APS k APS k PDP k
k

DDDPS, , P s , , , d
angular g delay yp power spectrum p

ADPS , DDDPS , , G MS d
l

APS APDS, d

angular power spectrum

P APSd
2008-11-20 Fredrik Tufvesson - ETI 085 5 2008-11-20

power

Fredrik Tufvesson - ETI 085

Angular dispersion
At the base station the angular spread is often modeled as L l i Laplacian 0 APS ( ) = exp( p( 2 )
S

Laplacian distribution, example

Angular spreads 5, 10, 20, 40 degrees
0.1

0.09

0.08

0.07

Typical rms angular spread:

Indoor office: 10-20 deg I d t i l 20 Industrial: 20-30 30 d deg Microcell 5-20 deg LOS, 10-40 deg NLOS Rural: 1-5 1 5 deg
pdf

0.06

0.05

0.04

0.03

0.02

0.01

0 200

150

100

50

50

100

150

200

Angle
2008-11-20 Fredrik Tufvesson - ETI 085 7 2008-11-20 Fredrik Tufvesson - ETI 085 8

Geometry-based stochastic channel models

Assign positions for scatterers according to given distributions Derive impulse response given the scatterers and distributions for the signal properties. Used in the COST 259 model the COST 273 model, model the 3GPP spatial channel model and the model, WINNER model

Channel measurements
In order to model the channel behavior we need to measure its properties
Time domain measurements
impulse i l sounder d correlative sounder

Frequency q y domain measurements

Vector network analyzer

Directional Di ti l measurements t

9

directional antennas real antenna arrays multiplexed arrays virtual arrays y

Fredrik Tufvesson - ETI 085 10

2008-11-20

Fredrik Tufvesson - ETI 085

2008-11-20

Impulse sounder
h meas t i, p ht i,
impulse response of sounder impulse response of channel

Correlative sounder
Transmit a pseudo-noise sequence and correlate with the same sequence at the receiver
Compare conventional CDMA systems Correlation peak for each delayed multipath component

p()

h( )

( ) h

-Tc Tc impulse response

Fredrik Tufvesson - ETI 085

correlation peak
2008-11-20 Fredrik Tufvesson - ETI 085 11 2008-11-20

measured impulse response

12

Frequency domain measurements

Use a vector network analyzer or similar to determine the t transfer f f function ti of f th the channel h l

Channel sounding directional antenna

Measure one impulse response for f each h antenna orientation

H meas ( f ) = H TXantenna ( f ) * H channel (f) TX h l ( f ) * H RXantenna RX

Time domain properties via FFT Using a large frequency band it is possible to get good time resolution As A f for ti time d domain i measurements, t we need dt to k know th the influence of the measurement system

2008-11-20

Fredrik Tufvesson - ETI 085

13

2008-11-20

Fredrik Tufvesson - ETI 085

14

Channel sounding antenna array

Measure one impulse response for each antenna element Ambiguity g y with linear array y

Real, multiplexed, and virtual arrays

Real array: simultaneous measurement at all antenna elements
linear array

RX

RX

RX

h()

h()

h()

h()

Multiplexed array: short time intervals between measurements at different elements Virtual array: long delay no problem with mutual coupling
15 2008-11-20 Fredrik Tufvesson - ETI 085

Digital Signal Processing

RX

x=0

x=d

x=2d Signal processing

x=(M-1)d

Digital Signal Processing

spatially resolved impulse response

RX

Digital Signal Processing

2008-11-20

Fredrik Tufvesson - ETI 085

16

Directional analysis
The DoA can, e.g., be estimated by correlating the received signals with steering vectors.
1 expjk 0 d cos a expj2k 0 d cos expjM 1k 0 d cos

High resolution algorithms

In order to get better angular resolution, other techniques f estimating for ti ti th the angles l are used, d e.g.:
MUSIC, subspace method using spectral search ESPRIT subspace method ESPRIT, MVM (Capons beamformer), rather easy spectral search method SAGE, , iterative maximum likelihood method

d sin

An element spacing of d=5.8 cm g of arrival of =20 and an angle degrees gives a time delay of 6.610-11 s between neighboring elements
Fredrik Tufvesson - ETI 085 17

Based on models for the propagation p , one measurement p point may y take 15 Rather complex, minutes on a decent computer

2008-11-20

2008-11-20

Fredrik Tufvesson - ETI 085

18

RUSK LUND, our broadband MIMO channel sounder

A fast switched measurement system for radio propagation investigations at 300 MHz MHz, 2 GHz and 5 GHz. Financed by Knut and Alice Wallenbergs stiftelse, FOI and LTH MIMO capacity limited by the switches, currently 32 elements at each side.

Its all about measuring g some delays... y

In MIMO systems we use the fact that there are several paths between the transmitter and receiver These paths are characterized by a
time delay, phase shift shift, attenuation, angle of departure and angle l of f arrival i l

The angle of departure and angle g of arrival result in a slight difference in time delay for each of the antenna elements

2008-11-20

Fredrik Tufvesson - ETI 085

19

2008-11-20

Fredrik Tufvesson - ETI 085

20

Its all about measuring g some delays... y

In practice we measure the transfer functions between each of f th the antenna t elements, l t and d we calculate l l t th the parameters t of interest

Working principle

Courtesy y MEDAV

2008-11-20

Fredrik Tufvesson - ETI 085

21

2008-11-20

Fredrik Tufvesson - ETI 085

22

Timing diagram
1 norm. .magnitude 0.8 0.6 0.4 0.2

Test signal g Multicarrier spread spectrum

Tx signalin frequency 1 norm. .magnitude 0.5 0 -0.5 Tx signalin time

0 51 5.1

5 15 5.2 5.15 52 5 25 5.25 frequency [GHz]

53 5.3

0.5 0 5 1 time [s]

1.5 1 5

MSSS - Test Squence

periodic broadband Signal high Correlation Gain low Crest Factor inherently band limited flexible in generation multiband possibility (Up- /Downlink)

2008-11-20

Fredrik Tufvesson - ETI 085

23

2008-11-20

Fredrik Tufvesson - ETI 085

24

The measurement system

200 kg of batteries to allow for 6 hours of mobile measurements 640 MHz sampling frequency frequency, to allow high Doppler frequencies separate ate PCs Cs to manage a age t the e data flow o from o t the e A/D / 2 sepa converters Oven controlled rubidium clocks to maintain synchronization during wireless measurements GPS and wheel sensors to position the system Broadband patch antennas with 128 antenna ports at 2.6 GHz Circular 300 MHz antennas with a diameter of 1.5 m
2008-11-20 Fredrik Tufvesson - ETI 085 25

RUSK LUND transmitter

bandwidths: up to 240 MHz frequency grid 10 MHz max. power 500 mW, with possibility for 10 W external amplifier lifi carrier frequency ranges
2200 2700 MHz, 5150 5750 MHz 235-387 MHz (20W)

Baseband ( (Arbitrary y Wave Form) ) Signal Generator Frequency Synthesizer Rubidium Reference M d l t Modulator Power Amplifier MIMO Control Unit GPS

Power Supply 24 V DC and 230 V AC

2008-11-20

Fredrik Tufvesson - ETI 085

26

RUSK LUND receiver

RF-Tuner RF Tuner High Speed ADC Automatic Gain Control (AGC) MIMO Control Unit Rubidium Reference High Speed Data Recorder 320 MByte/s, 500 GByte
Fredrik Tufvesson - ETI 085 27

Antennas
GPS Receiver Odometer Interface total amplification 72 dB AGC dynamic range 51 dB , adjustable dj t bl i in 3 dB steps, t intermediate frequency 160 MHz bandwidth 240 MHz Spurious free dynamic range 50 dB

To get good resolution we want large size arrays

4x16 dual polarized circular patch array

2008-11-20 Fredrik Tufvesson - ETI 085

4x8 dual polarized rectangular t l array

28

2008-11-20

Antennas cont.

RUSK LUND, Key Parameters

RF carrier frequency range
235-387 MHz 2200 2700 MHz, 5150 5750 MHz

Power: TX

RF carrier frequency grid: Measurement bandwidth up to 240 MHz (null-to-null bandwidth) p y MIMO capability:
1 MHz (300 MHz) (2 and 5 GHz) ) 10 MHz (

Antennas:

20 W (300 MHz) 500 mW and 10 W high power extension (2 and 5 GHz)

PDA device

300 MHz 7+1 circular sleeve antenna array laptop device

2008-11-20 Fredrik Tufvesson - ETI 085 29

16 TX antennas and 8 RX antennas (300 MHz) 32 TX antennas and 32 RX antennas simultaneously (2 and 5 GHZ)

7+1 circular monopole antenna array (300 MHz), 4x8 4 8 element l t planar l array, d dual l polarized (2 GHz) 4x16 element circular array, dual (2 GHz) ) polarized ( various application specific antennas

2008-11-20

Fredrik Tufvesson - ETI 085

30