38
03/25/22 1 4.3 Small Scale Path Measurements • multipath structure used to determine small- scale fading effects • Classification of Techniques for Wideband Channel Sounding (1) direct pulse (2) spread spectrum sliding correlator (3) swept frequency measurements

4.3 Small Scale Path Measurements

Embed Size (px)

DESCRIPTION

4.3 Small Scale Path Measurements multipath structure used to determine small-scale fading effects Classification of Techniques for Wideband Channel Sounding (1) direct pulse (2) spread spectrum sliding correlator (3) swept frequency measurements. T REP. Pulse Gen. f c. BW = 2/ T bb. - PowerPoint PPT Presentation

Citation preview

04/19/23

1

4.3 Small Scale Path Measurements

• multipath structure used to determine small-scale fading effects

• Classification of Techniques for Wideband Channel Sounding

(1) direct pulse

(2) spread spectrum sliding correlator

(3) swept frequency measurements

04/19/23

2

4.3.1: Direct RF Pulse System to measure channel impulse response

simple & cheap channel sounding approach - quickly determine PDP

• fundamentally a wide-band pulsed bistatic radar

• transmit probing pulse, p(t) with time duration = Tbb

• receiver uses wideband filter, BW = 2/ Tbb Hz

- envelope detector used to amplify & detect received signal

- results displayed or stored

Tbb = minimum resolvable delay between MPCs

e.g. let Tbb = 1ns BW = 2GHz & minimum resolvable delay = 1ns

BW = 2/TbbTbb

TREP

Detector StorageO-Scope

RxTx

PulseGen

fc

04/19/23

3

set o-scope to averaging mode system provides local average PDP

r(t)= )(

2

1 1

0i

N

i

iji tpe

direct pulse measurement yields immediate measure of |r(t)2|, where r(t) is given by

main problems:

• wide passband filter subject to interference & noise

• o-scope must trigger on 1st arriving signal, if 1st signal blocked or fades severely system may not trigger properly

• envelope detector doesn’t indicate phase of individual MPCs

- coherent detector would permit phase measurements

04/19/23

4

04/19/23

5

4.3.2 Spread Spectrum (SS) Sliding Correlator Sounding• probe signal is still wideband • possible to detect transmitted signal using narrowband receiver, preceded by wideband mixer

• improved dynamic range compared to pulsed RF system

SS: carrier PN sequence spreads signal over large bandwidth• Tc = chip duration• Rc = chip rate = Tc

-1

Tx

PN Gen

Tx Chip ClockRc = (Hz)

fc RxPNGen

Rx Chip Clock = β(Hz)

correlation BWBW2(-)

resolution Rc-1

(rms pulse width)BW2R

cwideband filter

Detectorat fc

StorageO-Scope

narrowband filter

System to Measure SS Channel Response

04/19/23

6

(1) SS signal generated by transmitter using some PN code

(2) received SS signal is filtered & despread using identical PN code

(3) sliding correlator implemented by using slightly slower chipping rate on receiver – causes periodic maximum correlation

(i) Tx PN Generator clock is slightly faster than Rx clock

(ii) when faster PN generator catches slower PN generator near identical alignment & maximal correlation

(iii) when two sequences are not maximally correlated • spread signal mixed with unsynchronized receiver chip sequence• signal is spread into bandwidth receivers reference PN sequence• narrowband filter following correlator rejects almost all incoming signal power

04/19/23

7

Sliding Correlator & SS approach enables receiver to • reject passband interference (advantage over RF pulse sounding)• realize significant processing gain (PG)

PG = (4.28) in

out

c

bb

bb

c

NS

NS

T

T

R

R

)/(

)/(22

(4.27)

null-to-null bandwidth given as:

BWnull = 2Rc

power spectrum envelope of transmitted signal given by

(4.26)

2

)(

)(sin

cc

cc

Tff

Tff

S(f) = = cc Tff 2Sa

04/19/23

8

(ii) different incoming multipaths have different delays - energy in individual paths will pass through correlator at different times- multipaths will maximally correlate at different times

(iii) after envelope detection - channel impulse response convolved with pulse shape of single chip is displayed on o-scope

For Sliding Correlator Rbb = -

Rbb = baseband information rate (Tbb = baseband information period)

- = frequency offset of transmit & receive PN clocks

(i) when incoming signal is correlated with receiver PN sequence

- signal collapses back into original bandwidth (despread)- the envelope is detected & displayed

04/19/23

9

Time Resolution of MPCs (width of excess delay bin) given by

= 2Tc = 2/Rc (4.29)

• if 2 MPCs are < 2Tc apart can’t be resolved

• minimal delay between resolvable MPCs = 2Tc

2Tc 1.5Tc

Sliding Correlation Process provides equivalent time measurements• updated each time 2 sequences are maximally correlated

Time Between Maximal Correlations is given by

T = Tc l (4.30)

Tc = chip period = Rc-1

= /- , slide factor (dimensionless)

l = 2n-1, chip sequence length (n bit m-sequence, )

Time Between Updates = 2 T

04/19/23

10

incoming signal is mixed on receiver with slower PN sequence• information transfer rate to o-scope = - - relative rate of 2 PN sequences

• signal essentially down-converted (collapsed) to low frequency, narrow band signal

- narrowband signal allows narrow band processing- eliminates passband noise & interference

• PG realized using narrowband filter with BW = 2(-)

• equivalent time measurements refer to relative times of MPCs as they are displayed on o-scope

• using sliding correlator, observed time scale on o-scope relates to actual propagation time scale

TimeObserved

actual propagation time = (4.33)

04/19/23

11

Time Dilation effect due to relative information transfer rate in sliding correlator

• Tc of 4.30 is observed time, not actual propagation time

• actual propagation delays are expanded by sliding correlator

• must ensure that PNseq > longest multipath delay

(4.34)PNseq = TclPN sequence period given by

estimated maximum unambiguous range of incoming MPCs is given by

PNseq · 3108m/s

04/19/23

12

• SS technique can reject passband noise – improving coverage range for given transmit power

• Sliding Correlator eliminates explicit Tx-Rx PN code synchronization

• However, measurements are not real-time, but derived as PN codes slide by each other - may require excessive time to measure PDP

04/19/23

13

4.3.3 Frequency Domain Channel Sounding

• vector network analyzer controls synthesized frequency sweeper

• S-parameter test-set monitors channel frequency response

• sweeper scans specified frequency band (centered on a carrier)

- steps through discrete frequencies

- number & spacing of discrete components affects resolution of impulse response measurement

Frequency Domain Channel Sounding System

RxTx

IFT

Vector Network Analyzer with Swept Frequency Oscillator

S-Parameter Test-Set

h(t) = F-1[H(w)]

S21(w) H(w) = Y(w)/X(w)

Y(w)port 2

X(w) port 1

04/19/23

14

For each frequency step the S-parameter test-set • transmits known signal on port 1

• monitors received signal on port 2

Network Analyzer processes signal levels to determine complex response of the channel over the measured frequency give as

S21(w) H(w)

- S21(w) = transmissivity

- transmissivity response is frequency domain representation of channel impulse response

- IFT used to convert back to time domain

Works well for short ranges if carefully calibrated & synchronized

04/19/23

15

4.4 Multipath Channel Parameters

• Power Delay Profile (PDP) is measured using techniques discussed in section 4.3

• several parameters are derived from PDP given in (4.18)

• represented as plots of relative received power as a function of excess delay with respect to fixed time delay reference

• average small-scale PDP found by averaging many samples of instantaneous PDP measured over local area

04/19/23

16

Spatial Separations of samples ¼ , depending on

(i) time resolution of probing pulse

(ii) type of multipath channels (indoor, outdoor,…)

e.g. at 2.4GHz = 125mm and ¼ 31mm 1.25 inches

Receiver Movement Ranges: range at which measurements will be consistent

• Indoor channels, 450MHz-6GHz range sample over receiver movement < 2m

• Outdoor channels sample over receiver movement < 6m

Small Scale Sampling must avoid large scale averaging bias in resulting small-scale statistics

04/19/23

17

Plots show typical PDP from outdoor & indoor channels determined from many closely sampled instantaneous profiles

1. Outdoor: 900MHz Cellular System worst-case in San Francisco• Display Threshold = -111.5 dBm per 40ns • RMS delay spread = 22.85us

0 10 20 30 40 50 60 70 80 90 100

-85-90-95

-100-105-110-115

Excess Delay Time (us)

Rec

eive

d S

igna

l Lev

el

(dB

m p

er 4

0ns)

04/19/23

18

2. Indoor: Grocercy Store at 4GHz • 39.4m path, • 18dB attenuation• 2mV/div, • 100ns/div• 51.7ns RMS• 43.0 dB loss

Excess Delay Time (ns)

Nor

mal

ized

Rec

eive

P

ower

(dB

)

-50 0 50 150 250 350 450

10

0

-10

-20

-30

04/19/23

19 15.4m LOS Distance NLOS Channel Response

18.5m LOS Distance LOS Channel Response

3. UWB Impulse Radio – Outdoor-Indutrial , Warren, MIfc = 4.4GHz, B-41dBm = 2GHz (3.1GHz-5.1 GHz)

mV

DC

mV

DC

04/19/23

20

Power Delay Profile used to determine multipath channel parameters• consecutive impulse response measurements collected & averaged over a local area

• averaged measurements based on temporal or spatial averages

• typically many measurements made at many local areas • enough to determine statistical range of multipath channel parameters for mobile system over large scale areas

4.4.1 Time Dispersion Parameters

Parameters that grossly quantify multipath channels are used to • develop general guidelines for wireless systems design• compare different multipath channels

Time-invariant Multipath PDP, P() derived from average of many snapshots of |hb(t,)|2 over local area

04/19/23

21

= mean excess delay

X = excess delay spread (X dB ) or maximum excess delay

= rms delay spread

Delays are measured relative to 1st detectable signal received at 0 =0

Eqns 4-35 thru 4-37 rely on relative amplitudes of MPCs within P() – not on absolute power level of P()

• commonly used to quantify time dispersive properties of wideband multipath channels

are defined from single PDP• and

• typical values for are us for outdoor & ns for indoor channel

Multipath channel parameters determined from PDP

04/19/23

22

kk

kkk

kk

kkk

P

P

)(

)(

2

2

(4.35)

= mean excess delay = 1st moment of PDP

= rms delay spread square root of 2nd central moment of PDP

kk

kkk

kk

kkk

P

P

)(

)( 2

2

22

2

where (4.37)

22 (4.36) =

X = maximum excess delay (X dB) of the power delay profile

• time delay during which multipath energy falls to X db below maximum (typically X = 10dB)

04/19/23

23

e.g. maximum excess delay X - 0

0 = 1st arriving signal

X = maximum delay at which multipath component is within XdB of strongest arriving multipath signal

• also called excess delay spread• always relevant to threshold relating multipath noise floor to maximum received multipath component

Delay measures, depend on selection of noise threshold

• noise threshold used in processing P() to differentiate between received MPCs and thermal noise

• if threshold set too low noise will be processed as multipath

• low threshold gives rise to artificially high delay measures

, 2,

04/19/23

24

indoor power delay profile

• maximum excess delay (X ) for MPCs within 10dB of maximum

• maximum excess delay defines temporal extent of multipath that is above a threshold

X = maximum excess delay = rms delay spread

= mean excess delay

= 45.05 ns

= 46.40 ns

X < 10dB = 84 ns

noise threshold = -20dB

Excess Delay (ns)

Nor

mal

Rec

eive

P

ower

(dB

)

10

0

-10

-20

-30-50 0 50 100 150 200 250 300 350 400 450

04/19/23

25

Environment Frequency max Notes Urban 910 MHz 1300ns avg

600 ns std-dev 3500nsNYC

Urban 892 MHz 10-25us - SF Suburban 910 MHz 200-310ns - Avg Typical Suburban 910 MHz 1960-2110ns - Avg Extreme Indoor 1500 MHz 10-50ns

median = 25ns- office bldg

Indoor 850 MHz - 270ns office bldg Indoor 1900 MHz 70-94ns avg 1470ns 3 SF bldgs

04/19/23

26

Mobile RF channel

• PDP & spectral response (magnitude of frequency response) are related by Fourier transform

• possible to obtain equivalent channel description in frequency domain using frequency response characteristics

Coherence Bandwidth, Bc

• analagous to delay spread parameters

• used to characterize channel in the frequency domain

• and Bc are inversely proportional, exact relationship depends on multipath structure

04/19/23

27

e.g. 4.4

(a ) Compute RMS delay spread for P()P()0dB

-10dB0 1us mean excess delay: us

2

1

)11(

)11()01(

rms delay spread:

us5.025.05.05.0 222

222

2

2

1

)11(

)11()01(us

(b) if BPSK used – what is Rb_max without equalizer (within Bc)

if Ts 5us Rs 200ksps and Rb 200kbps

ss

TT

1.0

1.0 for BPSK, normalized rms delay spread: d =

04/19/23

28

4.4.2 Coherence Bandwidth, Bc

Delay Spread is caused by reflected & scattered propagation paths

Bc is a defined relation derived from (rms delay spread)

• statistical measure of frequency range over which channel is considered flat• channel passes all spectral components with approximately equal gain & linear phase

e.g. frequency range over which 2 frequency components have strong potential for amplitude correlation

Consider 2 sinusoids with frequency f1 and f2 and fs = f2 – f1

- if fs > Bc signals are affected by channel very differently

- if fs < Bc signals are affected by channel nearly the same

Bc

Signal Level f

100%90%

Fading

04/19/23

29

Bc bandwidth related to frequency correlation function (FRC)

Bc FRC(50 )-1 FRC > 0.9(5 )-1 FRC > 0.5

• spectral analysis & simulation required to determine exact impact of multipath fading on particular signal• accurate multipath channel models are used in designing specific modems

estimated relationship between Bc &

FRC Bc

2us > 0.9 10KHz2us > 0.5 100KHz

20ns > 0.9 1MHz20ns > 0.5 10Mhz

larger delay spreads smaller coherence bandwidth

04/19/23

30

e.g. 4.4: determine X = maximum excess delay = rms delay spread = mean excess delay

0 1 2 3 4 5 (us)

Pr()0dB

-10dB

-20dB

-30dB

= (100)5 + (10-1)1 + (10-1)2 + (10-2)(0) = 4.38 us (100) + (10-1) + (10-1) + (10-2)

2 = (100)52 + (10-1)12 + (10-1)22 + (10-2)(0)2 = 21.07 us2

(100) + (10-1) + (10-1) + (10-2)

= 22 = 1.37 us

Bc = (5·1.37us)-1 = 146kHz (for FRC > 0.5)

•AMPS requires 30kHz bandwidth equalizer not required•GSM requires 200kHz bandwidth equalizer required

04/19/23

31

4.4.3 Doppler Spread and Coherence Time

Doppler Spectrum = received signal spectrum with range of fc fd

• fc = transmitted sinusoid wave

• fd = Doppler shift - function of relative velocity & angle of incidence

Doppler Spread, BD = measure of spectral broadening at receiver

• implies motion Doppler spectrum 0

• if baseband signal bandwidth, BS >> BD BD is negligible

Coherence Time, TC • characterizes time varying nature of channel’s frequency dispersion• time domain dual of BD and is inversely proportional to BD • statistical measure of interval when channel impulse response is invariant

- quantifies similarity of channel response at different times

- time interval when 2 signals have strong potential for amplitude correlation

04/19/23

32

> TC SB

1

• channel varies during baseband signal transmission• results in distortion at the receiver

If magnitude of baseband signal bandwidth < coherence time

• fm = v/ is maximum Doppler shift

TC

vfm

1 (4.40)a(i)

One measure of TC is given in terms of maximum Doppler shift

• assumes angle of incidence between Tx and Rx = 0

04/19/23

33

(ii) if TC is defined as interval when time correlation function > 0.5 then

(4.40b)TC mf16

9

v BS BS-1 fm TC

100m/s1GHz 1ns 0.3m

333Hz 537us10m/s 33.3Hz 5.37ms

100m/s10KHz 0.1ms 30000m

0.0033Hz 53.7 s10m/s 0.0003Hz 537s

e.g. assume v = 100m/s

04/19/23

34

• 4.40a = time duration when Rayleigh fading signal can have wide fluctuations

• 4.40b - often too restrictive

(iii) in Digital communications TC is often defined as geometric mean of 4.40a & 4.40b

(4.40c)TC mm ff

423.0

16

92

Definition of TC implies if 2 signals arrive at t1 & t2 with ts = t2-t1

if ts > TC both are affected differently by channel

if ts < TC both are affected approximately the same

04/19/23

35

Values for Tc

0

10

20

30

0 1 2 3 4 5

fm (KHz)

Tc

Tc1

Tc2

Tc3

04/19/23

36

• conservative value obtained from 4.40b TC = 2.2ms (454 Hz)-1

if RS ≥ 454 symbols/sec signal won’t distort from motion

• value from 4.44c TC = 6.77ms (150 Hz)-1

if RS > 150 symbols/sec signal won’t distort from motion

• any signal could still distort from multipath delay spread

e.g. v = 60mph (27.8 m/s) and fc = 900MHz

Determine distortion due to motion (i) determine TC from one of the equations

(ii) determine maximum symbol rate, RS for no distortion

RS >CT

1

04/19/23

37

e.g. 4.5: require that consecutive samples are highly correlated in time

• fc = 1900 MHz = 0.158m

• v = 50m/s• x = 10m is travel distance evaluated

Determine proper spatial sampling interval to make small-scale propagation measurements

• for high correlation in time, ensure sample interval = TC/2

- using conservative TC = 565usTC

mf16

9

- temporal sampling interval 282 us - spatial sampling interval: x = vTC/2 = 1.41cm

• number of samples over 10m = NX = 10/ x = 708 samples

• time required to make measurements = x/v = 0.2s• Doppler Spread: BD = fm = v/ = 316Hz

04/19/23

38

i. propagation effects (scattering, reflections) described by

• delay spread (e.g. )

• Bc = coherence bandwidth (spectral components affected the same)

ii. effects from motion of transceiver or objects described by

• Doppler spread, BD fm

• Coherence time, TC (temporal components affected the same)

Small-Scale time/frequency dispersive nature of RF channel

if frequency correlation > 90% then Bc ≈50

1

if time correlation > 50% then TC mf16

9