Submission doc.: IEEE 802.11-12/0416r1 Slide 1 Broadband Indoor TVWS Channel Measurement and Characterization at 670 MHz Date: 2012-03-14 Mar 2012 Ming-Tuo.

Submission

doc.: IEEE 802.11-12/0416r1

Slide 1

Broadband Indoor TVWS Channel Measurement and Characterization at 670 MHz

Date: 2012-03-14

Name Affiliations Address Phone email

Ming-Tuo Zhou NICT 20 Science Park Rd, #01-09/10 TeleTech Park, Singapore 117674

+65 67711007 [email protected]

Chunyi Song NICT 3-4, Hikarino Oka, Yokosuka, Kanagawa, 239-0847, Japan

[email protected]

Zhou Lan NICT 3-4, Hikarino Oka, Yokosuka, Kanagawa, 239-0847, Japan

[email protected]

Hiroshi Harada NICT 3-4, Hikarino Oka, Yokosuka, Kanagawa, 239-0847, Japan

[email protected]

Mar 2012

Ming-Tuo ZHOU, NICT

Submission

doc.: IEEE 802.11-12/0416r1

Abstract

An introduction to indoor TVWS channel measurement and results at 670 MHz

Slide 2

Mar 2012

Ming-Tuo ZHOU, NICT

Submission

doc.: IEEE 802.11-12/0416r1

Slide 3

Measurement Setup and Calibration

Measurement targets:Multipath delay spread, particularly, RMS delay spread

Channel impulse response

Path loss properties

Signaling: BPSK signal, 20 Mbps, 511-length pseudo random (PN) sequence, central frequency at

670 MHz

CW signal, central frequency at 670 MHz, for path loss measurement

Method:Custom designed receiver captures transmitted BPSK signal for measurement of

multipath delay spread and channel impulse response

For path loss measurement, the received signal power is measured by using a R&S FSU spectrum analyzer

Mar 2012

Ming-Tuo ZHOU, NICT

Submission

doc.: IEEE 802.11-12/0416r1

Measurement Setup and Calibration -- Instruments Used

Transmitter

Receiver & Spectrum Analyzer

Mar 2012

Ming-Tuo ZHOU, NICTSlide 4

Submission

doc.: IEEE 802.11-12/0416r1

Calibration Setup

Slide 5

Mar 2012

Ming-Tuo ZHOU, NICT

Submission

doc.: IEEE 802.11-12/0416r1

Measurement setup

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1

Measurement Parameters

Tx power 23 dBm

Cable loss 3.45 dB

Antenna Monopole type

Antenna gain 2.15 dBi

Channel Indoor LOS & NLOS

BPSK data rate 20 Mbps

BW 64 MHz (638-702 MHz,)

Center frequency 670 MHz

Tx antenna height 2.3 m / 2.6 m

Rx antenna height 0.7 m / 1.4 m

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1

Measurement Scenarios 1 – Office 1 – small size office/lab

• 185 sqm (=14.52 m×12.74 m)

• Includes staff room, 3 experiment rooms, director room, meeting room, kitchen, reception, etc

• Walls material: plywood/concrete

• Three transmitter antenna locations

• Tx antenna – 2.3m, Rx antenna – 0.7m

Mar 2012


Main door Experiment rooms

Meeting room Staff room

Reception

1 2 3

Tx-1

Tx-2

Tx-3

Submission

doc.: IEEE 802.11-12/0416r1

Measurement Scenarios 2 – Office 2 – medium size office/lab

• Includes two hall rooms, a conference room, several experiment rooms, a big lab room, store room, staff cubicle area

• Walls material: plywood/concrete

• One transmitter antenna position

• Tx antenna – 2.6m, Rx antenna – 1.4m

Mar 2012


4 m

Goods Lift

Hall Room 1Lab

Store Room

Office

Office

Hall Room 2

Office Spacewith staff cubics

WashRoom

ConferenceRoom

MeetingRoom

StairCase

Tx

Rx 1

Rx 6

Rx 7

Rx 2

Rx 3

Rx 4

Rx 5

Rx 8

Rx 17

Rx 15

Rx 16

Rx 14

Rx 18

Rx 9

Rx 10

Rx 13Rx 12

Rx 11

Room

Room Room

Room Room

RoomRoom Room

Submission

doc.: IEEE 802.11-12/0416r1

Some Pictures of Measurement Scenarios

Mar 2012


Office 1, staff room Office 1, meeting room

Office 2, Hall room 1 Office 2, office with staff cubicles

Submission

doc.: IEEE 802.11-12/0416r1

Extracting Rays

For each receiver location, BPSK signal is received and the normalized power delay profile is plot as function of time stamp

A peak detection method is used to extract rays. First, the calibrated Nyquist pulse is normalized to the peak of received

signal power delay profile, and then it is subtracted from the power delay profile

Second, the calibrated Nyquist pulse is normalized to the peak of the remain part, and then it is subtracted from the remain part of power delay profile

The above process is repeated until the peak power of the remain part is less than some threshold value, e.g., -30 dB

Each peak represents a received ray (path). Power (path gain) and relative time delay of each peak with higher power than threshold (e.g., -30dB) are recorded

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1

Extracting Rays (cont.)

Example

Mar 2012


Ray Delay (ns) 0.00 37.5 75.00 100 125 150 187.5 212.5

Ray relative power (dB) -26.73 -13.50 0 -2.14 -5.68 -11.11 -12.55 -21.41

0 100 200 300 400 500 600 700 8000

0.5

1R

ela

tive

po

wer

Relative time (ns)

0 100 200 300 400 500 600 700 800-100

-50

0

Re

lativ

e

pow

er (

dB

)

Relative time (ns)

0 100 200 300 400 500 600 700 800-30

-20-10

0

Relative time (ns)Re

lativ

e p

ow

er (

dB)

Submission

doc.: IEEE 802.11-12/0416r1

Extracting Rays (Cont.)

The method extracting rays is verified by experiment of reconstructing the signal.

Signal is reconstructed by summing copies of calibrated Nyquist pulses, each of which is weighted by a ray power (path gain) and delayed with ray delay

The reconstructed signal is close to the measured one, as illustrated by following example

0 100 200 300 400 500 600 700 800-50

-40

-30

-20

-10

0

Time (ns)

Re

lativ

e p

ow

er (

dB)

MeasuredReconstructed

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1

RMS Delay Spread

RMS delay at each receiver location is calculated for each measurement scenario

Median RMS delay is then calculated for each scenario (with different threshold)

Scenario LOS (ns)

(-30dB) LOS (ns)

(-20dB) LOS (ns) (-10dB)

NLOS (ns) (-30dB)

NLOS (ns) (-20dB)

NLOS (ns) (-10dB)

Office 1 Tx-1 32.42 31.33 20.88 41.09 36.20 27.44

Office 1 Tx-2 20.20 18.96 14.54 36.09 34.43 18.24

Office 1 Tx-3 20.51 18.72 10.34 32.37 30.36 26.48

Median for office 1 20.51 18.96 14.54 36.09 34.43 26.48

Office 2 38.52 34.67 19.04 42.47 39.32 28.19Overall Median 26.47 25.15 16.79 38.59 35.32 26.96

Mar 2012

Slide 14 Ming-Tuo ZHOU, NICT

Submission

doc.: IEEE 802.11-12/0416r1

Channel Modeling

We observed that rays arrive in clusters usually.

Clusters may be formed by super-structures and walls inside buildings. In this study, super-structures may be metal cabinet or goods lift, and so on

Clusters attenuate exponentially, because of propagation delay

0 100 200 300 400 500 600 700 8000

0.5

1

Re

lativ

e p

ow

erRelative time (ns)

0 100 200 300 400 500 600 700 800-100

-50

0R

ela

tive

p

ower

(d

B)

Relative time (ns)

0 100 200 300 400 500 600 700 800-30

-20

-10

0


lativ

e p

ow

er (

dB)

0 100 200 300 400 500 600 700 8000

0.5

1

Re

lativ

e p

ow

er

Relative time (ns)

0 100 200 300 400 500 600 700 800-100

-50

0

Re

lativ

e

pow

er (

dB

)

Relative time (ns)

0 100 200 300 400 500 600 700 800-30

-20-10

0


lativ

e p

ow

er (

dB)

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1

Channel Modeling (cont.)

Comparison to S-V model

Mar 2012


New findings in this study Saleh-Valenzuela (S-V) model

A cluster may appear before the strongest cluster in some cases

The most left-side cluster is the strongest

Inside a cluster rays power may increase with time delay

Inside a cluster, rays power attenuate with time delay

Submission

doc.: IEEE 802.11-12/0416r1


A qualitative explain to the above new findings is that different rays/clusters may have different antenna gain, due to difference in angle-of-arrival (AoA)

Lately arrived rays may have larger antenna gain, because of larger AoA

Then although they have larger propagation delay, their power are larger than earlier rays

An earlier cluster may have averagely smaller antenna gain than a later cluster, then later cluster may be stronger

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1


Illustration of qualitative explanation for rays arrival earlier may be weaker in power

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1


Proposed low-pass impulse response modelRays arrive in clusters, each cluster may consist of a group of rays.

At middle is the strongest cluster, with cluster arrival time of 0

At the middle of each cluster is the strongest ray and it represents the cluster arrival time Tl, which is arrival time relative to the strongest cluster

Rays arrival time relative to the strongest ray is , where l is the cluster index, m is the ray index

Cluster arrival time is modeled as Poisson arrival process with fixed arrival constant. Left side and right side clusters may have different cluster arrival time

Rays arrival time is modeled as Poisson arrival process with fixed arrival constant, too

Clusters decay exponentially with cluster arrival time, on both left and right sides (with possible different decay constant)

Rays decay with ray arrival time

lm

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1


Illustration of the proposed low-pass complex channel impulse response model (with comparison to S-V model)

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1


Mathematically, the low-pass complex channel impulse response model is given by

202

20

, 0

, 0

lm

lm

llm

l

e m

e m

0

0

2002

02

00

, 0

, 0

l

l

T

lT

e l

e l

l

lm

l

MLj

lm l lml L m M

h t e t T

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1


Parameters extracted from measurementsNumber of clusters: one left-side clusters occasionally, 2-4 right-side clusters

Number of rays: 2-5 left-side rays in the strongest clusters, median of 0.67 for other clusters

Cluster decay constant: left-side clusters 200ns to 1000ns, right-clusters: 25ns – 50 ns (median 33ns)

Ray decay parameter: ranges between 16ns to 22ns if treat as constant, in small room or NLOS cases, rays decay faster

Cluster arrival rate: both left-side and right-side clusters may have similar arrival rate, around 1/60ns

Ray arrival rate: in LOS cases ray arrival rate is 1/12.5ns usually, NLOS case has smaller ray arrival rate (1/25ns, even 1/37.5ns)

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1

Path Loss Properties

Radio signals attenuate with distance exponentially

For each measurement distance, we took median power of the received samples

By fitting the path loss with minimum mean square error (MMSE), the exponential constant of LOS and NLOS in Office 1 is 2.02 and 2.09, respectively, indicating a strong waveguide effect

In Office 2, LOS and NLOS has exponential constant of 3.16 and 3.56, respectively.

Mar 2012


Submission

doc.: IEEE 802.11-12/0416r1

Conclusion

Indoor channels at TVWS frequency of 670 MHz in both small- and medium-size mixed office/lab environment are measured and characterized

(Overall) median of RMS delay spread for LOS with -30dB threshold is 26. 47ns, and is 38.59ns for NLOS with -30dB cutoff threshold. With smaller threshold, RMS delay spreads are smaller

A low-pass complex impulse response model is proposed based on classic Saleh-Valenzuela (S-V) model and new findings

Path loss constants are presented

Mar 2012


Submission doc.: IEEE 802.11-12/0416r1 Slide 1 Broadband Indoor TVWS Channel Measurement and Characterization at 670 MHz Date: 2012-03-14 Mar 2012 Ming-Tuo.

Documents

mingtuo zhou

signal power delay profile

measurement of multipath

received signal power

store room

director room

conference room

mincludes staff room