Doc.: IEEE 802.22-05/0107r3 Submission January 2006 Soo-Young Chang & Jianwei Zhang, Huawei...

January 2006

Soo-Young Chang & Jianwei Zhang, Huawei Technologies Slide 1

doc.: IEEE 802.22-05/0107r3

Submission

WAVEFORM MODULATED WRAN SYSTEMIEEE P802.22 Wireless RANs Date: 2006-1-16

Name Company Address Phone email Soo-Young Chang

Huawei Technologies

6000 J Street, Dept EEE, Sacramento, CA 95819-6019

916 278 6568 [email protected]

Jianwei Zhang Huawei Technologies

No. 98, Lane 91, Eshan Road, Pudong, Pudong Lujiazui Software Park, Shanghai, China 200127

86-21-68644808-24638

[email protected]

Authors:

Notice: This document has been prepared to assist IEEE 802.22. 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 contributor grants a free, irrevocable license to the IEEE to incorporate material contained in this contribution, and any modifications thereof, in the creation of an IEEE Standards publication; to copyright in the IEEE’s name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEE’s sole discretion to permit others to reproduce in whole or in part the resulting IEEE Standards publication. The contributor also acknowledges and accepts that this contribution may be made public by IEEE 802.22.

Patent Policy and Procedures: The contributor is familiar with the IEEE 802 Patent Policy and Procedures http://standards.ieee.org/guides/bylaws/sb-bylaws.pdf including the statement "IEEE standards may include the known use of patent(s), including patent applications, provided the IEEE receives assurance from the patent holder or applicant with respect to patents essential for compliance with both mandatory and optional portions of the standard." Early disclosure to the Working Group of patent information that might be relevant to the standard is essential to reduce the possibility for delays in the development process and increase the likelihood that the draft publication will be approved for publication. Please notify the Chair Carl R. Stevenson as early as possible, in written or electronic form, if patented technology (or technology under patent application) might be incorporated into a draft standard being developed within the IEEE 802.22 Working Group. If you have questions, contact the IEEE Patent Committee Administrator at [email protected].>

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

Abstract

• In this proposal, a system concept is suggested for the IEEE802.22 WRAN standard. A set of waveforms are suggested for WRAN systems. In these systems, one TV channel frequency band is divided into 16 subbands each of which has its own waveform. In the time domain, these waveforms are added and transmitted. Multiple access schemes are suggested by applying orthogonal codes in the frequency domain.

• These waveforms are generated by utilizing full digital processing in this proposal.

• Sensing and dynamic frequency selection (DFS) schemes using FFTs are proposed.

• These schemes are evaluated by simulation.

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

CONTENTS

• Introduction

• Overview of the system concepts

• Waveform of a subband

• Modulations

• Multiple access

• Sensing of incumbent signals

• Dynamic frequency selection (DFS)

• Performance evaluation

• Conclusions

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

INTRODUCTION

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

INTRODUCTION/BACKGROUND

• Use short duration waveforms: processed purely in the time domain, not in frequency domain– Simple concept: only a few components in TX and RX– Simple digital processing Low complexity Low cost– No components for processing frequency information except LNA

and wideband tuning (e.g. filter, osc., etc.)– Excellent co-existence capability due to adaptive frequency band

use – flexible to eliminate forbidden bands (e.g. active incumbent TV user bands, active microphone bands, etc.) • Dynamically frequency bands can be assigned to CPEs

• New waveforms have steep out-of-band rejection around the edges of the band.

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

BACKGROUND

• Spectrum usage of TV broadcast industries– the average TV market in the United States uses approximately 7 high-

power channels of the 67 that it is allocated. This leaves an abundance of free channels that could be used for wireless access.

– With both the House and the Senate having recently passed bills requiring television broadcasts to switch from analog to digital sometime in early 2009, the 700-MHz band (channels 52 to 69) will be cleared of programming and moved to lower frequencies (channels 2 to 51). The 700-MHz band will be set aside for public-safety emergency transponders and for bidding by wireless networks.

in this proposal only channels 2 to 51 are considered.• Three possible ways suggested in one article to protect

interference with incumbent users– Listen-Before-Talk (LBT)– Geolocation/Database: GPS receivers installed in CPEs– Local beacon: locally transmitted signal used to identify incumbent users

Unused Digital TV Channels Could Increase U.S. Wireless Access, Federal action could allow unused channels at lower frequencies to be used for unlicensed wireless networks, Eric S. Crouch, Medill News Service, PC World, Saturday, November 19, 2005, http://www.washingtonpost.com/wp-dyn/content/article/2005/11/18/AR2005111800083_pf.html

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

CHANNEL AVAILABILITY

• Questioned whether there will be significant channel availability for unlicensed use in major urban areas during the DTV transition.– There is likely to be substantial channel availability during transition.– The issue of channel availability during the DTV transition is likely to be

short-lived.– In rural areas, there is spectrum available now and there will be for the

foreseeable future.

• Bill Rose’s email to 22 email reflector, Wed, November 23, 2005 10:05 am – “The analysis shows that even in congested markets like Dallas/Ft. Worth,

40 percent of the TV channel spectrum will remain unused after America's DTV transition. In more rural markets like Juneau, Alaska, as much as 74 percent will be available.”

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

INTERFERENCE WITH INCUMBENT USERS

• 73 million TV sets

• DTV disruption issue

• Public safety interference

• Newsgathering and sports programming production

• Interference with theaters, churches, and school events

• Cable services

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

PLAUSIBLE MITHS

• Myth 1– ‘Digital implementation needs more complexity and is not easily realizable with existing

technologies.’ Digital implementation can be realized with less complexity and simple hardware and provide

full flexibility and adaptivity. As processing power increases and technologies advance, full digital processing is the trend.

• Myth 2– ‘Lower frequency is not easy to manage or implement.’ Unless high transmit power is not considered, digital processing method can be easily applied

for lower frequency band without using more complex algorithms. 4 W EIRP can be handled without difficulties.

• Myth 3– ‘Since this technology was not realizable yesterday, today also it is not easy to realize.’ Since technologies advance rapidly, more sophisticated and conceptual ideas should be

realized in the near future and considered for future applications. Moore’s law says that processing power increases double every 18 months: cost and

complexity can be decreased with the same rate.

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

WM AND OFDM HAVE SAME ORIGIN

• Almost the same theory and very simple concepts used for WM* and OFDM– Orthogonality of frequency components used: refer to the next slide

– Major parameters from the same formula used in the same way

• Sampling rate, symbol duration, frequency separation, number of samples, sampling interval, etc

T00

t

T

F0

f

F0

Using Discrete Fourier Transform

* WM: waveform modulation

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

ORTHOGONALITY OF SINUSIODS

• A key property of sinusids is that they are orthogonal at different frequencies. That is, • This is true whether they are complex or real, and whatever amplitude and phase they may have. All that matters is that the frequencies be different. Note, however, that the sinusoidal durations must be infinity. • For length N sampled sinusoidal signal segments exact orthogonality holds only for the hamonics of the sampling rate-divided-by-N , i.e., only for the frequencies

• These are the only frequencies that have a whole number of periods in samples• Ex. N=100 for 4 ns pulse duration, fs=25 GHz fk=k*25*10**9/100=2.5*10**8*k=0.25*k GHz

For any integer k, fk can be determined center frequencies of each subband can be determined

•http://ccrma.stanford.edu/~jos/r320/Orthogonality_Sinusoids.html

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

WHY WM RATHER THAN OFDM?

• WM has strong points in– WM has almost a flat spectrum for each frequency segment while OFDM has a sync shape

spectrum if the same parameters applied WM has flatter spectra inside band and more suppression out of band WM does not use FFT/IFFT while it has to use de-emphasis at receiver de-emphasis means a different value for each sampled component which is stored in memory at receiver no burden for de-emphasis

WM OFDM

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

WHY WM RATHER THAN OFDM?

• WM has strong points in– WM covers signals of whole TV band from 0 to the highest band while OFDM does

for one TV band WM does not need up/down conversion while OFDM needs it OFDM needs additional hardware (or another signal processing branch) to cover two bands simultaneously

f

whole target frequency band frequency band which can be covered by

WM without additional hardware

frequency band which can be covered by

OFDM without additional hardware

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

DISADVANTAGES OF WM OVER OTHER PROPOSED OFDM

• High sampling rates required– Better spectrum using sync shape

envelopes in the time domain relieves high sampling rate restraints

• Vulnerable to synchronization errors– Sync shape envelope helps to attain

accurate synchronization– WM: 180 samples/2.7 us

OFDM: 2k samples/330 us: 10 times narrower sample interval for WM than for OFDM because WM uses much less samples (or subbands)

Zero crossing helps to attain accurate synchronization

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

ADVANTAGES OF WM OVER OTHER PROPOSED OFDM

• Flexibility in band selection, band expansion/reduction/elimination – similar to CDMA concepts to

superpose waveform signals in a symbol duration

– Other signals look like noise to the receiver

• No need of up/down conversion– Simple hardware to

implement

+

+

+

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

CLASSIFICATION OF CONCEPTS (1)

Up/down conversion

• Using up/down conversion: other OFDMs and DFS– Lower sampling rates

– Less vulnerable (or more margin) to synchronization errors

• Without using up/down conversion: WM and new OFDM– Flexible to frequency band use in expansion/reduction/elimination

– Easy to have sensing and DFS

January 2006


doc.: IEEE 802.22-05/0107r3

Submission


Signal processing for single band or multi bands

• Single band can be covered: other OFDM– More efficient if FFT/IFFT inputs/outputs used– One band sensing and DFS

• Multi bands can be covered: WM, new OFDM, DFS– Flexible to use of frequency bands:

expansion/reduction/elimination without additional hardware: covers all bands

– DFS and sensing easily implemented for all bands

January 2006


doc.: IEEE 802.22-05/0107r3

Submission


Waveform modulated or using FFTs/IFFTs

• WM based– needs de-emphasis at receiver

– x2 or x4 more sample durations and numbers of samples needed than non-WM

– Does not need FFTs/IFFTs

• Non-WM using FFT/IFFT– More difficult to cover broadband with single FFT/IFFT

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

OVERVIW OF THE SYSTEM CONCEPTS

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

KEY COMPONENTS

• Modulation

• Source coding

• Channel coding/error control– FEC and ARQ

• Interleaving

• Waveform generation

• Antenna

• Multiple access

• Dynamic frequency band allocation

• Synchronization

• LNA

• Message relaying: repeaters

• Sensing of incumbent user signals

• Dynamic frequency selection (DFS)

• Transmit power control (TPC)

• Detection

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

KEY SYSTEM CONSIDERATIONS

• Simple design to implement– Adopt full digital implementation: ~33 Msamples/sec DACs/ADCs needed

• More radiated power efficient – Almost flat spectrum inside the assigned band

• Use almost full bands assigned to the users– Easy to meet out-of-band requirements (or frequency mask)

• Need less bits to represent samples for equal out-of-band suppression than other concepts– Achieve maximum range coverage

• More system flexibility in real time– More flexible to various applications and requirements– More dynamically adaptive to available frequency bands in real-time basis

• 6, 7, or 8 MHz BW• Single band or multiple bands which are adjacent to or depart from each other

– Scalable information rate adaptive and dedicated to applications• Basic physics says (square root of coverage) x (number of simultaneous users) x (information rate/user) = constant

– High value of this constant can be achieved• Scalable with coverage, number of users, information rate

– For. Ex: wide coverage with lower information rate or vice versa

• Adaptively balancing between uplink and downlink information rates– Basically 1.5 Mbps max for downlink and 384 Kbps for uplink required– Adaptively balancing: symmetric and asymmetric according to specific applications

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

FREQUENCY PLAN

• Flexible enough to satisfy any frequency band given and to avoid any forbidden bands 6, 7, or 8 MHz bandwidth can be easily adopted A part of a TV channel band can be eliminated for Part 74 device services pulse waveforms can be adaptively tailored to any frequency mask or

band applied with any forbidden bands• With any given frequency band, the whole frequency band can be

used to enjoy more transmitted power and achieve higher data rates. Due to steep suppression around the edges of the band 3.8 dB more power used than Gaussian pulse’s case for the same

frequency band 3.8 dB more margin for link budget

One TV channel is divided into 16 subbands – 4 groups/channel and 4 subbands/group

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

FREQUENCY SUBBANDS• One TV channel frequency band is divided into 4 groups• Each group has 4 subbands

– BW of a subband = 6 MHz /16 = 0.375 MHz – Each subband has its own waveform: base waveform– If a part of a given band should be abandoned – e.g., due to active microphone

operation - one or more of corresponding subbands can be eliminated.

f

subband 1 subband 2 subband 3 subband 4

f

group 1 group 2 group 3 group 4

x MHz x+6 MHz

w21 w22 w23 w24base waveform

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

WAVEFORM OF A SUBBAND

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

BASE WAVEFORM OF A SUBBAND

• Base waveform shaping– From mathematical derivation/expression

• Almost sync function shape envelope

– Shape: duration: 5.4 us• The required symbol duration to maintain the orthogonality is 2.7 us.

• Two times the required symbol duration is used to make more brick-wall like spectrum.

• Longer waveform durations make more immune to inter-symbol interference (ISI) due to comparably small delay spreads.

– Spectrum: almost flat throughout the whole band• Flatness depends on the number of samples: more samples/waveform

entails flatter spectrum and more suppression out of the band

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

WAVEFORMS FOR ONE TV CHANNEL

• Base waveform– For each subband, there is one base waveform which has flat

spectrum almost throughout the subband as shown in the next slides.

• Group waveforms– Group i has four base waveforms:

• wi1, wi2 , wi3 , and wi4

– Group i has a set of waveforms which are combinations of four base waveforms:

mij,=a* wi1 +b* wi2 +c* wi3 +d* wi4

where a, b, c, and d are determined by modulation method applied

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

BASE WAVEFORM (EX2)

• The base waveform below – for carrier frequency = 3.5 MHz, bandwidth = 0.469 MHz,

sampling rate = 10 samples/us

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

SPECTRAL FLATNESS vs NO OF

SAMPLES/WAVEFORM

• With the same waveform, spectral flatness and out-of-band suppression depend on the number of samples for the waveform duration

– More samples makes the spectrum flatter: flatter inside the band and more suppression outside the band

– Power ratio=power with perfectly flat spectrum / power with less perfectly flat spectrum inside the band

– For the cases• Bandwidth = 0.469 MHz• Pulse width = 9 us• No. bits/sample = 8• No. samples/waveform = 90 and 180

2 3 4 5 6 7

-14

-12

-10

-8

-6

-4

-2

0

2

frequency

ampl

itude

in d

B

Frequency Masked Frequency domain spectra (GROUP 4)

2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

-14

-12

-10

-8

-6

-4

-2

0

2

time

ampl

itude

in d

B

Frequency Masked Frequency domain spectra (GROUP 1)

No. samples/waveform =

90

No. samples/waveform =

180

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

GENERATION OF BASE WAVEFORM

• How can waveforms be generated – Digital way?

base waveform information is stored in ROM continuous signals of waveforms are generated by DACs at TX can be generated with relatively lower sampling rate DACs

• 180 samples/waveform used for this proposal• 2x4=8 base waveforms/group for binary representation: 8x4=32 base waveforms per TV

channel64x4=256 base waveforms/group for 64-ary representation: 256x4=1024 base waveforms per TV channel

• Max 180x1024=184,320 sample information stored in ROM per TV channel 184 Kbytes ROM per TV channel needed to store waveform information if 8 bits/sample is adopted: 184 K x 100 = 18 Mbytes ROM for 100 TV channels

• Waveforms are generated using DACs which have a sampling rate of 33.4 Msamples/sec. – 180 samples / 5.4 us = 33.4 Msamples/sec

– Analog way?• No idea

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

BASE WAVEFORMS OF ONE GROUP

• For four subbands – assuming each subband has 1 MHz BW

– If smaller BW, larger pulse width

x 1 2 3 4 5 6 7 8 9 10 +x0

0.20.40.60.81

1.21.41.61.82

Frequency( MHz.)

amplitude

t (us)0 4

+

+

+

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

BASE WAVEFORMS OF ONE GROUP

For four subbands - for smaller BW, larger pulse widthFor another example, for a case of BW = 0.469 MHz

subband 1

subband 2

subband 3

subband 4

us

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

SPECTRUM OF MULTIPLE SUBBANDS

For no. of samples per waveform: 10

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

ORTHOGONALITY OF WAVEFORMS

• For each subband, one base waveform exists – 16 base waveforms throughout whole band (or four groups):

w11(t), w12(t), w13(t), w14(t), w21(t), . . . . , w43(t), w44(t)

– Each waveform is almost orthogonal to each other or perfectly orthogonal after de-emphasis at RX

• Each group has – 2**4=16 waveforms for binary base waveform modulation (BPSK) or – 4**4=256 waveforms for quaternary base waveform modulation (QPSK)– 16**4=256K waveforms for quaternary base waveform modulation (16QAM)– 64**4=16M waveforms for quaternary base waveform modulation (64QAM)– These waveforms are orthogonal to each other after de-emphasis at RX

mij,=a* wi1 +b* wi2 +c* wi3 +d* wi4

where a, b, c, and d are complex numbers determined by modulation method appliedfor BPSK a, b, c, and d are +1 or -1for QPSK a, b, c, and d are +1, +j, -j or -1for nQAM a, b, c, and d are complex numbers

• m1,1= -w1 - w2 – w3 – w4, . . . . , m4,16= w13+ w14+ w15+ w16 for BPSK

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

CORRELATIONS BETWEEN WAVEFORMS

• Correlation

where : kth sample of ith base waveform of a group for N samples/waveform

• Ratio of correlations = autocorrelation/crosscorrelation for various N values• Orthogonality holds for sinusoidal waveforms with some conditions (Orthogonality condition,

refer to next slide), but the waveforms used here are not sinusoidal with a fixed envelope– At receiver, simple de-emphasis can be used to make pure sinusoidal for a period

• mij*mij=(a* wi1 +b* wi2 +c* wi3 +d* wi4 )(a* wi1 +b* wi2 +c* wi3 +d* wi4) where mij is the waveform transmitted and mij is the waveform generated at RX after de-emphasis

• After integration for one waveform duration, only autocorrelation terms remain• Orthogonality can hold at receiver during detection for matched waveforms

– What is the best sampling frequency such that orthogonality can be achievable?• Less than 8 bits/sample will be enough for orthogonality evaluation? – needs to be verified

– “Power consumption of ADCs goes up exponentially with resolution”, EE times, Jan 17, 2005, pp 49

)()( *

1

kskslationcrosscorre j

N

ki

2

1

*

1

|)(|)()( ksksksationautocorrelN

kii

N

ki

)(ksi

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

ORTHOGONALITY OF SINUSIODS

• A key property of sinusids is that they are orthogonal at different frequencies. That is, • This is true whether they are complex or real, and whatever amplitude and phase they may have. All that matters is that the frequencies be different. Note, however, that the sinusoidal durations must be infinity. • For length N sampled sinusoidal signal segments exact orthogonality holds only for the hamonics of the sampling rate-divided-by-N , i.e., only for the frequencies

• These are the only frequencies that have a whole number of periods in samples• Ex. N=100 for 4 ns pulse duration, fs=25 GHz fk=k*25*10**9/100=2.5*10**8*k=0.25*k GHz

For any integer k, fk can be determined center frequencies of each subband can be determined

•http://ccrma.stanford.edu/~jos/r320/Orthogonality_Sinusoids.html

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

ONE EXAMPLE FOR PREVIOUS SLIDE

• Frequency band: 500-506 MHz, 6 MHz Bandwidth– Whole band is divided into 16 subbands– Each subband has 0.375 MHz bandwidth– 16 Carrier frequencies: 500.1875, 500.5625, . . . , 505.8125 MHz– Frequency separation = h* fs / N = 0.375 MHz where h is an

arbitrary positive integer– fs = N / T = 0.375*N*10**6/h where T is a waveform duration

• T = h / 0.375 * 10**(-6) = h /0.375 us• For ex. for h=1, T=2.7 us, for h=3, T=8 us, for h=4, T=10.7 us, . . .

– For ex., fs=698 MHz• Frequency separation=fs/N=3/8 MHz for h=1• N=698*8/3=1862 ~ 2K points/waveform 2K samples/waveform

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

CORRELATIONS BETWEEN TWO BASE WAVEFORMS

correlation

correlation ratio

w11 w12 w13 w14

w11 0.020984

1/1

0.0012155

17.264/9.7396

2.2562×10-5 930.05/3957.3

3.4173×10-6

6140.6/9681.8

w12 0.0012155

17.264/9.7396

0.020984

1/1

6.8651×10-6

305.66/106.69

2.2562×10-5

930.05/3957.3

w13 2.2562×10-5

930.05/3957.3

6.8651×10-6

305.66/106.69

0.020984

1/1

0.0012155

17.264/9.7396

w14 3.4173×10-6

6140.6/9681.8

2.2562×10-5

930.05/3957.3

0.0012155

17.264/9.7396

0.020984

1/1

– # of samples = 180 # of samples = 90

– Correlation ratio = autocorrelation/crosscorrelation

– Correlations totally depend on the number of samples used.

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

MODULATIONS

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

MODULATION/MULTIPLE ACCESS EFFICIANCY

• Energy or power efficient? joule/sec– Energy=power*time– Power limited by spectral mask and EIRP

• Pmax=PSD/MHz*BW

to use more energy, more time needed to be transmitted totally related to transmit time for WRAN, BW~6MHz short duration waveforms can be used for higher data rates one possibility to increase energy by using multiple pulses for one bit (or symbol) need to use more power under frequency mask to have higher power power constrained with frequency mask and EIRP for WRAN case new waveforms needed to fit the frequency mask to have more transmitted power

• Spectrally efficient? bit/Hz– limited bandwidth given– More complex modulation schemes have to be applied entails higher system complexity

• Time efficient? bit/sec– For higher rate, more important : needs a short duration waveform for one symbol Needs to put more information in a symbol duration Needs more sophisticated modulations

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

POSSIBLE MODULATIONS FOR EACH WAVEFORM

• Each waveform can be modulated by using the following modulation schemes depending on required data rates, system complexity, detection method, etc

Modulation No. of levels Complexity Data rate Detection method

OOK 2 (+1, 0) lowest low Non-coherent/coherent

Anti-podal: BPSK 2 (+1, -1) low low Coherent/differential

QPSK 4 moderate moderate Coherent/differential

n level mod n high high Coherent/differential

nQAM n high high Coherent/differential

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

SPECTRUM OF BASE WAVEFORM

-10 -8 -6 -4 -2 0 2 4 6 8 10-30

-20

-10

0

10

Frequency

ampl

itude

in d

B

PSD representation in dB scale.

-10 -8 -6 -4 -2 0 2 4 6 8 10-15

-10

-5

0

5

10

Frequency

ampl

itude

in d

B

Integration of PSD for spectral Flatness (log scale)

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

EXAMPLES OF WAVEFORMS OF A GROUP (BPSK)

Submission

m1,1(t) m1,11(t) m1,16(t)

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

SIGNAL FOR RANDOMLY GENERATED 10,000 BITS

0 50 100 150 200 250 300 350 400 450 500-0.1

-0.08

-0.06

-0.04

-0.02

0

0.02

0.04

0.06

0.08

0.1

time in ns.

ampl

itude

time domain spectra for 10000 randomly generated bits.Total Waveforms=2500

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

SPECTRUM FOR SIMULATED SIGNAL

For a signal of the previous slide

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

TRANSMITTER STRUCTURE

• Simple structure with full digital processing concept– FEC encoder

– Interleaver

– Waveform generator

– Modulator

– Antenna

Data manipulator

modulator

waveform generator

Data in

antenna

source codingchannel codinginterleaving

This part can be realized using digital processing

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

TRANSMITTER BLOCK DIAGRAM

ROM, group 1

ROM, group 2

ROM, group 3

ROM, group 4

DAC

DAC

DAC

DAC

waveform transformer




data manipulator

S/P converter

encodinginterleavingencryption

input data

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

RECEIVER STRUCTURE

• Simple receiver structure– Antenna - Waveform generator

– LNA - Demodulator

– Data detector

– De-interleaver

– Channel decoder

– Synchronizer

demodulatordata

de-manipulator

waveformgenerator

synchinformation

retriever

antenna

LNA

Data out

detectorwide band tuner

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

RECEIVING BLOCK

delayblock

ADC correlator

ROM

LNA

Sampling rate ~33.4 MHz

de-emphasis waveform

information stored

Wideband tuner

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

MULTIPLE ACCESS

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

DOWNLINK/UPLINK DUPLEXING

• Code division duplex (CDD)– Full duplex mode

– Provide full flexibility in uplink/downlink balancing• Symmetric/asymmetric

– More time efficient than TDD and more spectrally efficient than FDD

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

MULTIPLE ACCESS (MA)

• Possible MAs considered– Frequency hopping (FH)

among subbands/groups• Not efficient because of

higher system complexity and less usage of power

– TDMA• Less time efficient• More difficult to synchronize

– Direct-sequence (DS) CDMA• Less time efficient and more

complex to process

– FDMA/OFDMA• More complex

• New MA needed?

f

t

Group 1

Group 2

Group 3

Group 4

16 frequency bins

time domain bins

t4t2 t3t1 t5

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

MAPPING FREQUENCY BINS TO WALSH ENCODED SYMBOLS

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

PROPOSED MULTIPLE ACCESS

• An orthogonal set of 8 8-bit Walsh codes is used– Max autocorrelation, zero crosscorrelation each other– One code consists of 8 frequency domain bins– Minimal Hamming distance of this code set is 4 – mostly 8

• One frequency bin error can be corrected while three bin errors can be detected; works like an ECC code; increases robustness

• 64 simultaneously operated users– For one user, two Walsh codes (16 bits) are assigned– One time domain bin is occupied by two codes

• two codes represent one symbol; one time domain bin represents one symbol; one time domain bin deliver one symbol

• Hamming distances between two user codes are 4 and mostly 8.

• For each frequency bin waveform, BPSK, QPSK, 16QAM or 64QAM is applied according to signal environments – or according to the distance between a CPE and the base station.

• Use of full frequency band– Codes are spread over the full band – entails higher power efficiency

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

DUPLEXING/MULTIPLE ACCESS

• Duplexing– Code division duplexing (CDD)

• Downlink MA– Orthogonal code assigned with sectorization

• Walsh codes MA

• Antenna sectorization: 3 sectors with 120 degrees coverage for each sector

downlink channel capacity increased by three times

• Uplink MA– Orthogonal code assigned with omni directional antennas

uplink channel capacity increased by three times due to sectorization

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

SENSING OF INCUMBENT SIGNALS

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

SPECTRA OF TV CHANNELS

Analyzing the Signal Quality of NTSC and ATSC Television RF Signals.htm, Glen Kropuenske, Sencore

NTSC and DTV signal spectra

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

NTSC TELEVISION BAND

Conventional Analog Television - An Introduction

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

DTV PILOT FREQUENCY

Conventional Analog Television - An IntroductionPresented at the IEEE Broadcast Technical Society 49th Symposium September 24, 1999

Henry Fries and Brett JenkinsThales Broadcast & Multimedia, Inc.

Southwick, MA

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

DTV SIGNAL VIEWED ON A SPECTRUM ANALYZER



Southwick, MA

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

DTV OUT-OF-BAND“SHOULDERS”



Southwick, MA

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

VSB TV PARAMETERS

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

TV CHANNELS IN U.S.

• Currently with 6 MHz bandwidth for each channel,– VHF low band: Chs 2-6 54-88 MHz

– VHF high band: Chs 7-13 174-216 MHz

– UHF band: Chs 14-69 470-806 MHz *

• After DTV transition,– VHF low band: Chs 2-6 54-88 MHz

– VHF high band: Chs 7-13 174-216 MHz

– UHF band: Chs 14-51 470-698 MHz *

• In this proposal, channels after DTV transition are considered.– Enough channels are expected to be maintained for WRAN.

• For other bandwidths – 7 and 8 MHz – the system concept can also be applied by changing system parameters.

* Ch 37 is reserved for radio astronomy

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

METHOD 1 (1) SENSING INCUMBENT SIGNALS

• TV band signal sensing for one channel band– Use only spectral components – not time domain components

• Less sensitive on other parameters used to design TV band tuners – for example, Phase noise, etc.

– Use FFT transform of received TV band signals at the receiver for only one TV band

• After wide band tuning and down converting or down converting and low pass filtering

• BW=F=6 MHz• Sampling interval T=1/B=1/6 us, sampling rate=BW=6 MHz• Frequency resolution (or frequency separation) F0=3 KHz• Time period T0=1/F0=1/3 ms• Number of samples needed N0=T0/T= 2 KHz• Needs 2K point FFT

January 2006


doc.: IEEE 802.22-05/0107r3

Submission


T00

t

T

F0

f

F0

Sense antenna

LNAcos2fptwhere fp: left edge freq. of the channel

LPF ADC FFT detector

Discrete Fourier Transform

Sense Receiver Structure

January 2006


doc.: IEEE 802.22-05/0107r3

Submission


• Sensing procedure for TV signals– Several frequency components taken in a 6 MHz band

• F50, F103, F200, F417, and F800

– Compare these values• Correlation method: compare the shape of spectrum of received signals

– Calculate correlations with pre-stored values for NTSC and DTV signals

– If one of these values is larger than predetermined values, the judgment is that NTSC or DTV signal exists.

• Pilot detection method: check whether a pilot signal exists– Calculate the ratio of pilot component to another component

– If F417/F1200 > thn, this signal is NTSC

– If F103/F1200 > thd, this signal is DTV

– Repeat this procedure for several symbol periods to make sure the sensing results

January 2006


doc.: IEEE 802.22-05/0107r3

Submission


• Sensing procedure for wireless microphone signals– Two types of wireless microphone systems according to frequency usage

• Single frequency systems• Frequency agile systems

– Wireless systems should NOT be operated on the same frequency as a local TV station.

• Only open (unoccupied) frequencies should be used. In the U.S., each major city has different local TV stations.

– Microphone signal detection procedure: sensing the spectral components using FFT devices

• For every 3 KHz in a 6 MHz band a spectral component is measured and compared with other components.

• If considerable components in a 200 KHz band exist, a wireless microphone is considered to be operated in that band: if consecutive six components have considerable amount of energy, a microphone signal is detected.

January 2006


doc.: IEEE 802.22-05/0107r3

Submission


• After DTV transition in the U.S.,– VHF low band: Chs 2-6 54-88 MHz– VHF high band: Chs 7-13 174-216 MHz– UHF band: Chs 14-51 470-698 MHz *

• n consecutive bands in VHF High or UHF band selected for WRAN services

– The whole band of n bands is divided into n*l subbands• Each band has l subbands; each subband has 6000/l KHz bandwidth

– At receiver, the received signal after down conversion is inputted to a l*n point FFT• By comparing FFT output signals, currently operated incumbent users can be identified and

categorized – NTSC, DTV, or Part 74 devices

• With method all incumbent signal throughout the whole band (n TV bands) can be detected simultaneously – Periodically all CPEs and BSs can do this sensing to update the list of

active incumbent users

* Ch 37 is reserved for radio astronomy

January 2006


doc.: IEEE 802.22-05/0107r3

Submission


• NTSC signal sensing– After down conversion with (fp+1.25) MHz frequency shift, the received signal is

inputted to l*n point FFT devices– Compare the FFT outputs

• DTV signal sensing– After down conversion with (fp+0.30944) MHz frequency shift, the received signal

is inputted to l*n point FFT devices– Compare the FFT outputs

• Part 74 device sensing– After down conversion with fp MHz frequency shift, the received signal is inputted

to l*n point FFT devices– Compare the FFT outputs

• Various comparison methods can be considered– Correlation method or pilot detection method used in Method 1 is suggested for TV

signals – If some consecutive strong components in 200 KHz exist, Part 74 device is

considered to operate in this band.

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

DYNAMIC FREQENCY SELECTION (DFS)

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

DYNAMIC FREQUENCY SELECTION (DFS)

• Select k consecutive bands out of n bands– Each band is divided into l subbands.– Each subband in selected bands carries information.

– One complex number of information is assigned to each subband • For k selected bands, one complex number is assigned to each subband

– A complex number is determined by the constellation which depends on modulation adopted for the system:

• BPSK: two points: can deliver only one bit per symbol duration

• QPSK: four points: can deliver two bits per symbol duration• 2m QAM: 2m points: can deliver m bits per symbol duration

• For (n-k) not-selected bands, zero is assigned to each subband: – By assigning zero to each subband and applying OOK, no signal is transmitted

– Dynamic frequency selection can be achieved: by changing the assigned values for subbands of a unused band dynamically, WRAN service band(s) can be selected.

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

DYNAMIC FREQUENCY SELECTION (DFS)

• Select k consecutive bands out of n bands

f

Band 0 Band 1 Band k-1Selected bands

subband 0subband 1

subband 2

subband l-1

WRAN Incumbent user WRANWRAN/incumbent

1 0 1Assigned data

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

PROPOSED RECEIVER STRUCTURE

• At receiver, data receiving and incumbent signal sensing are executed simultaneously.– Without having separate receiving and processing branches– Using sensing method 2– If more precise sensing is needed, sensing method 1 may be applied with

an additional signal processing block – needs one more ADC and FFT.

receive antenna

LNAcos2fptwhere fp: left edge freq. of the channel (or whole target band)

LPF ADC FFT

detector

demod

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

ONE EXAMPLE

• For n=32, k=4, l=60 – Four bands of 6 MHz BW each selected out of thirty three bands assigned

for WRAN– Each band is divided into 60 subbands

• each subband has 100 KHz bandwidth

– FFT parameters• Frequency separation F0=100 KHz• Symbol duration T0=1/100 KHz=10 us• Sampling rate F=6x32=192 MHz• Sampling interval T=1/F=1/192 us• No. of samples in a symbol duration N=T0/T=1920• 2048 point FFT/IFFT can be used

– Dynamically the system can select any four consecutive bands out of 32 bands or easily expand the operating band in these four bands.

• In this proposal, l=16.

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

SELECTION OF BANDS

• n: total number of TV channel bands observed– Total target frequency band for which incumbent signals can be sensed– In this band, some TV channel bands can be occupied to send information for

WRAN services– As n increases, total band increases and number of samples of FFT increases:

number of points of FFT?IFFT increases– Sampling interval is inversely proportional to n: sampling interval is inversely

proportional to n)• k: number of bands used to deliver information

– k determines symbol rates: for a 6 MHz channel, its symbol rate is 6 Msymbols/sec determines data rates which depends on modulation

– As k increases, bandwidth used for information delivery increases and the number of inputs and outputs of FFT used for information delivery increases.

• l: number of subbands in a TV channel band– l determines one FFT symbol duration– As l increases, frequency separation decreases and one symbol duration increases

• The number of points of FFT/IFFF is proportional to n*l

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

FREQUENCY BAND STRUCTURE

f

k bandsn bands

one TV bandl subbands

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

PERFORMANCE EVALUATION

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

DATA RATES

• Assumptions– 64 simultaneously operated CPEs and a base station per each sector– One TV channel available (6MHz BW)

• Aggregated data rates estimated per TV channel – BPSK applied for a waveform

• 1 bit/waveform x 1/5.4 waveforms/us x 64 users = 11.85 Mbps– QPSK applied for a waveform

• 2 bit/waveform x 1/5.4 waveforms/us x 64 users = 23.7 Mbps– 16 QAM applied for a waveform

• 4 bit/waveform x 1/5.4 waveforms/us x 64 users = 47.41 Mbps– 64 QAM applied for a waveform

• 6 bit/waveform x 1/5.4 waveforms/us x 64 users = 71.11 Mbps• Data rates per subscriber

– Min 11.85 Mbps/64 = 0.19 Mbps, Max 71.11 Mbps/64 = 1.08 Mbps, both ways symmetrical

• Spectral efficiency– Min 11.85 Mbps/6 MHz= 1.98 bits/sec/Hz, Max 71.11 Mbps/6 MHz = 11.85

bits/sec/Hz

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

COMPLEXITIES

Scheme Waveform modulated OFDM

Sampling rate 33.4 MHz 6 MHz

FFT/IFFT Not required One IFFT at TX

One FFT at RX

Up/down conversion Not required Up conversion at TX

Down conversion at RX

No. of calculation for one symbol detection

(multiplication/division only)

180 samples x 64 = 11.5K multiplications at

RX

2K x 2K = 4M multiplications at RX

Memory space to store waveform information

18 Mbytes for 64 QAM and 100 TV channel

band

Not required

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

CONCLUSIONS

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

SUMMARIZED FEATURES

• With 6 MHz bandwidth– One band is divided into 16 subbands– Each subband has its waveform: all 16 waveforms are nearly orthogonal to each other– Modulations: BPSK, QPSK, 16QAM, and 64QAM– Sectorization can be implemented with directional antennas for BSs and omni directional

antennas for CPEs to increase data rates• Exactly same hardware for CPEs as that without sectorization• Three receiving blocks needed for a base station with derectional antenna

– Multiple access: coded MA in frequency domain• 64 users at one instant

– Aggregated data rates: min 11.85 Mbps, max 71.11 Mbps– Data rates per subscriber: min 0.19 Mbps, max 1.08 Mbps– Spectral efficiency : min 1.98 bits/sec/Hz, Max 11.85 bits/sec/Hz

• Advantages over other OFDM concepts– Simpler concept: much simpler implementation/lower complexity than other competing

technologies– Pure digital implementation– More flexibility in scalability, up/down balancing– Realized with lower sampling rate DACs and ADCs

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

WHY THIS PROPOSAL?

• Very simple concepts / architecture– Directly generated pulse waveforms using ROMs– Processing in digital methods

• No need to have analog devices (e.g., mixer, LO, integrator, etc) except LNAs low complexity / low cost / low power consumption

• High out-of-band rejection with equal complexity– More transmit power and more bandwidth efficient high data rates can be achieved

• High adaptability to frequency, data rate, transmit power requirements high scalability in frequency band, data rate, system configuration, uplink/downlink balancing, waveform, etc.

• More transmit power used under frequency mask– More margin in link budget: 3.8 dB more by using full power under any frequency-

power constraints with waveforms adaptive to frequency mask Spectrally efficient / more received signal power More chance to intercept signals

January 2006


doc.: IEEE 802.22-05/0107r3

Submission

References

1. “Power consumption of ADCs goes up exponentially with resolution”, EE times, Jan 17, 2005, pp 49

2. http://ccrma.stanford.edu/~jos/r320/Orthogonality_Sinusoids.html

Doc.: IEEE 802.22-05/0107r3 Submission January 2006 Soo-Young Chang & Jianwei Zhang, Huawei...

Documents

Transcript of Doc.: IEEE 802.22-05/0107r3 Submission January 2006 Soo-Young Chang & Jianwei Zhang, Huawei...