November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)

Submission Title: [UWB Channel Modeling Contribution from CEA-LETI and STMicroelectronics]Date Submitted: [28 October, 2002]Source: [J.Keignart, JB.Pierrot, N.Daniele] Company [CEA-LETI] [Ph.Rouzet] Company [STM]Address [CEA-Grenoble, 17 rue de Martyrs 38054 Grenoble cedex 9, FRANCE]Voice [+33 4 38 78 58 11], Fax [+33 4 38 78 51 59] STMicroelectronics 39 Chemin du Champ des Filles 1228 Geneve Plan-les-Ouates, SwitzerlandVoice [+41 22 929 58 66], Fax [+41 22 929 29 70]E-Mail:[julien.keignart@cea.fr][Philippe.Rouzet@st.com]Re: []Abstract: [This contribution proposes a UWB path loss and multipath model for assisting in the evaluation of possible UWB physical layer submissions for a high-rate extension to IEEE 802.15.3. ]Purpose: [In this presentation, we propose Saleh-Valenzuela (S-V) model parameters extracted from an office environment measurement campaign performed by CEA-LETI. A Channel Impulse Response (CIR) processing method, not yet implemented is also proposed at theoretical level. This method allows to process all the points of the CIR without applying a threshold on the level of the paths.]Notice: This document has been prepared to assist the IEEE P802.15. 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 contributors acknowledge and accepts that this contribution becomes the property of IEEE and may be made publicly available by P802.15.

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

UWB Channel Modeling Contribution from CEA-LETI and STMicroelectronics

Julien Keignart, Norbert DanieleJean-Benoît Pierrot

CEA-LETI(CEA/ Direction de la Recherche Technologique)

Philippe RouzetSTMicroelectronics – Advanced System Technology Division

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Overview • Measurement setup• Measurement environment• Frequency domain results• Time domain results• Some parameters for a S-V channel model• Proposal of a method for CIR processing

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Measurement Setup (1)• Vector Network Analyzer: Agilent PNA-E8357A

Bandwidth range: 2GHz to 6GHz with 1601 points• Conical monopole antennas: ARA CMA-118/A• Power Amplifier: AMF-3D-0580-20P Miteq• LNA: AFS3-00100800-32-LN & AFS4-00100800-45-8P Miteq• Measurement remotely controlled by LabVIEW

®

• Post-processing performed by using Matlab®

• Calibration performed in an anechoic room with 1m reference separation (obviously including antennas)

• Measurement from 3 (lab.) or 9 (flat) successive averages (static environment during recording)

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Measurement Setup (2)• Complex frequency response is recorded

– Impulse response obtain from Matlab IFFT• Variable attenuator from 0dB to 50dB for high and low

antenna separation compensated for post-processing Tx (Port 1)

VectorNetworkAnalyzerAgilentE8357A

Rx (Port 2)

PA

LNA LNA HPF Rx Ant.

Tx Ant.Post-processingAnalysisin Matlab

CMA-118/AAMF25dB

Attenuator

0 to 50dB

PC withLabVIEW 6i

Channel

CMA-118/AFc=2GHzAFS329dB

AFS435dB

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Measurement environment (1)• 3 different configurations:

– LOS– NLOS: with at least one plasterboard wall– NLOS²: with at least one concrete wall

• Antenna separations up to 20m (common height of 125cm)• Indoor office/laboratory environment up to 20m (447 CIR):

– LOS (17 different positions) (1-6m)– NLOS (74 different positions) (3-20m)– NLOS² (58 different positions) (2-20m) (partially analyzed)

• Indoor flat environment up to 17m (2430 CIR):– LOS (116 different positions) (1-8m)– NLOS (45 different positions) (9-13m)– NLOS² (109 different positions) (7-17m) (partially analyzed)

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Measurement environment (2)

Office/laboratory map

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Frequency domain measurement results (1)• An ideal free space (FS) path (no ground reflection, no

multipath) has a path loss that is proportional to the square (=2) of the separation d, and with λ the wavelength:

where is the path loss exponent and c is a power scaling constant included in calibration

• Friis formula suggest a 1m path loss equal to 44.5 dB at 4GHz center frequency.

• Attenuation has been directly calculated from the total received power in frequency domain.

cdddPLdB

1010 log104log10)(

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Frequency domain measurement results (2)

1 2 3 4 5 6-15

-10

-5

0LOS - Path Loss VS Distance

Tx-Rx Distance (m)

Pat

h Lo

ss (d

B)

2 3 4 5 6 10 20-55

-50

-45

-40

-35

-30

-25

-20

-15

-10NLOS - Path Loss VS Distance

Tx-Rx Distance (m)

Pat

h Lo

ss (d

B)

• Path loss results for office/laboratory environment

• LOS path loss factor: = - 1.6319• NLOS path loss factor: = - 3.6822 • NLOS² path loss factor: = - 4.7759

2 4 6 8 10 14 18-70

-60

-50

-40

-30

-20

-10NLOS² - Path Loss VS Distance

Tx-Rx Distance (m)

Pat

h Lo

ss (d

B)

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Frequency domain measurement results (3)• Path loss results for flat environment

• LOS path loss factor: = - 1.6655• NLOS path loss factor: = - 4.9657• NLOS² path loss factor: = - 7.2422

10-29

-27

-25

-23

-21

-19

NLOS - Path Loss VS Distance

Tx-Rx Distance (m)

Pat

h Lo

ss (d

B)

8 9 141 2 3 4 5 6 7 8 9

-16

-12

-8

-4

0

LOS - Path Loss VS Distance

Tx-Rx Distance (m)

Pat

h Lo

ss (d

B)

6 7 8 9 10 20-50

-45

-40

-35

-30

-25

-20

-15NLOS² - Path Loss VS Distance

Tx-Rx Distance (m)

Pat

h Lo

ss (d

B)

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Post-processing generalities• The arrival time of the first path can be found by using a

threshold from the peak in LOS and NLOS cases. But in NLOS² the exact distance seems to be necessary!

0 50 100 150 200 250 300 350 400-80

-70

-60

-50

-40

-30

-20

-10

0

Time (ns)

Am

plitu

de (d

B)

Example of channel impulse response

10dB of the peak16dB of the peak for the arrival time

Noise floor

10dB above noise floor

Number of multipath Mean Excess Delay

&

RMS Delay Spread

The value of 16dB has been found by minimizing the error between the exact distance measured and the calculated distance from threshold.

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Time domain measurement results (1)• Three parameters from channel impulse response are analysed

for a bin size of 250ps obtain from complex baseband ifft:

Environment Mean Excess Delay (ns)

RMS Delay Spread (ns)

Number of paths within 10dB

LOS (0-4m)Flat 6.53 11.45 3.4

Lab/office 6.42 10.07 2NLOS(4-10m) Lab/office 16.01 14.78 46.8

NLOS(10-20m) Lab/office 18.85 17.64 75.8

– Number of paths within 10dB corresponds to multipath with an amplitude higher than 10dB under the maximum path.

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Time domain measurement results (2)• Comparison of path amplitude (linear) distribution between two

common models:– Rayleigh– Log-normal

• The goodness of fit obtained is evaluated by mean of Kolmogorov-Smirnov test with 1% significance level (only from laboratory measurement):

Pass Rate ofLog-normal (%)

Pass Rate of Rayleigh (%)

LOS (flat) 76 63

NLOS (lab/office) 97 76

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Time domain measurement results (3)• Standard deviation of log-normal variable in LOS and NLOS

cases:– Decreasing with Excess Delay

0 10 20 30 40 50 60 70 80 90 1000

1

2

3

4

5

6

Std

of L

og a

mpl

itude

(dB

)

Excess Delay (ns)

NLOS – Laboratory/office

0 5 10 15 20 25 30 35 400

1

2

3

4

5

6

Std

of L

og a

mpl

itude

(dB

)

Excess Delay (ns)

LOS – Flat

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Elements for a Channel Model (1)• Firstly, we need to insist on the fact that these results need to

be validated with more measurements

• Saleh-Valenzuela seems to be a good channel model:Model parameters: = cluster arrival rate

= ray arrival rate = cluster decay factor = ray decay factor

and = standard deviation of log-normal fading term (dB)

Adel A. M. Saleh, Fellow, IEEE, and Reinaldo A. Valenzuela, Member, IEEE, “A Statistical Model for Indoor Multipath Propagation”, IEEE Journal on selected areas in communications, Vol. SAC-5, No. 2, February 1987

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

Elements for a Channel Model (2)• It is possible to get this model parameters using brute

force method to match with the three parameters obtained from measurement.

LOS (0-4m) NLOS(4-10m) NLOS(10-20m)Flat Lab/office Lab/office Lab/office

(1/nsec) 0.007 0.0065 0.09 0.17 (1/nsec) 1.27 1.26 1.05 1.06 (10-9) 30 29 19.7 19.8 (10-9) 10 10 1.06 1.1

(dB) Decrease with excess delay5.5 to 1 (see previous slide)

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

CIR Processing Method Proposal(1)Now, a new method will be presented. The main advantage of this one compared to conventional approach is the fact that there is no more need of detecting paths using a threshold. Consequently, ALL multipaths are included in model parameters! • Diversity of environment not only one kind of distribution

• Well Estimate distribution needs plenty of measure

A priori knowledges:

• Hypothesis on distribution:– Ray density (Poisson, Double poisson [Saleh-Valenzuela, Δ-K model,…])– Ray energy (Rayleigh, log-normal,…)

• Estimate distribution parameters (needs less measure)

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

CIR Processing Method (2)

Standard channel: l

Ttj

lt l

leh +ttr

thtx

t iid zero meanl U[0 2π]

t

2nP

Energy measure (bins size:Δ):

1 *2 n

n ttn dtrrP

ttt hr with:

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

CIR Processing Method (3)Estimate the energy mean:

20

1

22 NEPE

nTnl

ln

l

Ray delays and amplitudes distributions hypothesis for this example:

t

ll

l elttTP

!1

1

2

2

22

2

21 ReR

P

• Delay Poisson:

• Amplitude Rayleigh:

• Mean amplitude dependent to ray delay: s

ssll tatTE 2

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

CIR Processing Method (4)

1

2

nTnl

l

l

E

Theoretical steps:

l

lll nTndP2

1;22

l

ll

TE

lll

T

ll

nTnTdPTdP 1;

2

2

22

ls

ls

k

nkksk

s eennlskl

a1

0

1!1!!1

And: )(,,,, 0102 nSnn fNaafPE

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

CIR Processing Method (5)

Measurement steps:

• M measures in each bin: nnm

nPfPEmP

Mn

221ˆ 2

• One estimator per bin:

2

120

ˆˆNPP

T

• Estimate parameters using non-linear optimisation method:

1

0

21

20

ˆ

ˆ

NP

P

f

f

N

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

CIR Processing Method (6)

Conclusion:• This method can be generalized with others assumptions on models

- Amplitude as log-normal- Mean as exponential- Delay as double-Poisson (Saleh-Valenzuela, Δ-K model,…)- …

• It doesn’t need any particular detection (spike detection method,…)

November 2002

CEA-Leti / STM

doc.: IEEE 802.15-02/444

Submission

ConclusionMost of the parameters are in good accordance with the S-V model (IEEE802.15-02/279r0 & 02/368r2) proposed by Intel.

Several measurements have been performed out of the IEEE scope, for example in typical European domestic environment with an antenna separation higher than 10m.

Need to agree on a common post-processing method like in slide titled “Post-processing generalities”.

Proposal of a CIR processing method• Software implementation planed for the next months• Test and comparison on the CIR measured