Hopping localized transmission to improve UL transmit power Document Number: S80216m-08/171 Date Submitted: 2008-03-10 Source: Tom HarelVoice:+972-3-9207175.

Hopping localized transmission to improve UL transmit powerDocument Number: S80216m-08/171

Date Submitted: 2008-03-10

Source:Tom Harel Voice: +972-3-9207175

E-mail: [email protected] Lomnitz E-mail: [email protected]

Intel CorporationVenue: IEEE session #54, Orlando, FL.Base Contribution:

None.Purpose:

For discussionNotice:

This document does not represent the agreed views of the IEEE 802.16 Working Group or any of its subgroups. It represents only the views of the participants listed in the “Source(s)” field above. It is offered as a basis for discussion. It is not binding on the contributor(s), who 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 802.16.

Patent Policy:The contributor is familiar with the IEEE-SA Patent Policy and Procedures:

<http://standards.ieee.org/guides/bylaws/sect6-7.html#6> and <http://standards.ieee.org/guides/opman/sect6.html#6.3>.Further information is located at <http://standards.ieee.org/board/pat/pat-material.html> and <http://standards.ieee.org/board/pat >.

2

Introduction• This contribution proposes a way to

improve UL link budget by improvement of the TX power

• The information is provided for discussion only, as preparation for UL symbol structure discussions expected in next IEEE session

3

Link budget issue in the uplink• From 802.16m SRD, section 7.4, Cell coverage:

– “the link budget of the limiting link (e.g. DL MAP, UL bearer) of IEEE 802.16m shall be improved by at least 3 dB compared to the WirelessMAN-OFDMA Reference System.”

• One of the factors affecting UL link budget is the transmit power• Mobile TX power is limited due to the following factors:

– High PAPR - large variation of the of OFDM signal envelope– Non-linear “practical” power amplifier– Constraints

• Out of band emission is limited by spectral mask (varies by regulation)• Minimum EVM is needed (in-band noise limitation), dending on MCS• PA may have power consumption limitation (in addition to peak power

limitation)• 802.16e OFDMA uplink performance is limited with respect to the

downlink (TX power 23 dBm vs. 46 dBm, while maximum sub-channelization gain ~12 dB)

• Maximum TX power (of lowest rate) is limited by spectral mask requirement (since EVM requirement loosens for low rates)

4

PAPR reduction methods• PAPR reduction techniques improve peak power

• The actual performance gain from PAPR reduction methods like tone-reservation, tone-injection etc. is very small

• Reasons:– Improving the peak power doesn’t have a 1:1 impact on the maximum TX power:

• It has small effect on OOB and in-band distortion since most of them created by non-peak signal

• EVM and OOB improvement relates in a ratio of approx 1:3 to TX power improvement (in dB)

• For example ideally limiting the OFDM amplitude to 7dB has ~0.5dB gain in TX power (depending on model and mask)

– These methods insert some overhead or loss in performance that balances some of the gain

• Clipping & filtering is an effective method to be applied in the transmitter and no standardization is needed for it, except correct definition of the EVM levels

• We propose to further improve the maximum TX power not by changing the signal amplitude distribution but by different use of the spectrum

=> “PAPR reduction” methods evaluation

5

Facing the spectral mask –localized-OFDMA

• Non-linear PA causes spectral expansion of the transmitted signal. Narrower signal’s spectrum will cause narrower expansion.

• We suggest to allocate narrow localized chunk of subcarriers for power limited users

• This simple mechanism has very good performance, although it doesn’t change the signal’s PAPR.

Original OFDM signal with OOB

Narrow band signal with same spectral density

Narrow band signal amplified to meet spectral mask requirement

6

Localized-OFDMAThe following results show the gain obtained with actual OFDM signal and the following parameters:

PA model: RAPP-3

OFDM parameters: 10Mhz, FFT1024, wideband=PUSC 3 subchannels, narrowband = 72 subcarriers

Mask: FCC & HUMAN

-30 -20 -10 0 10 20 30-110

-100

-90

-80

-70

-60

-50

-40

-30

-20

Frequency [MHz]

x

x(f)

[dB

m/H

z]

Spectral density and masks. TX power = 30.53

-30 -20 -10 0 10 20 30-110

-100

-90

-80

-70

-60

-50

-40

-30

Frequency [MHz]

x

x(f)

[dB

m/H

z]

Spectral density and masks. TX power = 23.30

-30 -20 -10 0 10 20 30-110

-100

-90

-80

-70

-60

-50

-40

-30

Frequency [MHz]

x

x(f)

[dB

m/H

z]

Spectral density and masks. TX power = 25.15

23.3 dBm

25.2 dBm30.5 dBm

Wide-band OFDM signal

Narrow-band OFDM signal, band edgeNarrow-band OFDM signal, band center

+7dB

+2d

B

=> Spectral efficiency of localized OFDMA

7

Adding frequency diversity by hopping

• For high mobility user the frequency diversity gain in MIMO 2x2 is ~6dB (PUSC versus AMC)

• In localized transmission we lose this diversity gain• To combine the frequency diversity of UL-PUSC with

power advantage of localized OFDMA, fast frequency hoping should be applied (e.g. hop duration of 2 symbols), therefore we propose hopping localized transmission

• On the other hand hopping localized requires continuous chunk of spectrum to be allocated to a single user which poses a limit on other users.

• Therefore we propose to limit this type of allocation to cell-edge (power and throughput limited) users

8

Hopping localized allocations

• We propose that a mix of three allocation types will be supported by the UL symbol structure:– Power limited diversity users: hoping localized (HL) allocation– Closed loop (low mobility) users: constant localized allocation

(“AMC”)– High throughput diversity users: distributed allocation (similar to

UL-”PUSC”)

• The power boosting in HL allocation can be a function of the location in the band (maximum power can be applied to ~80% of the band, lower power in the edges)(See slide Localized gain as function of location in the band)

Time

Freq

Dwell time

4 hops

Min

imum

allo

catio

nw

idth

(#t

one)

9

Dwell time tradeoffs• The basic allocation unit is a time-frequency rectangle. It’s size is

affected by:– Large number of sub-carriers reduces the maximum sub-channelization

gain, therefore span maximum time (e.g. 2 subframes) minimum frequency

– Given the frequency width, the tradeoff on dwell time:• Small dwell time => more hops, more diversity• Large dwell time => higher pilot efficiency

• Recommended parameters:– A hop per 2 symbols yields an optimum point between pilot loss and

diversity loss, assuming TTI=2 subframes– Having 6 hops within a frame yields reasonable frequency diversity

(assuming interleaving over time). – Assuming UL transmission may span TTI=2 subframes, we assume 3

hops per subframe, i.e. hop every 2 symbols– This yields a tile of e.g. 9x2, 12x2 or 18x2 which has reasonable pilot

efficiency

10

Backup

11

Methodology of transmission methods comparison

• Methodology– Generate sample signals– Compress them in PA model– Measure spectrum and EVM– Estimate performance and compare different

methods:• Maximum TX power• EVM dependence on TX power• SE versus link margin

12

Allocation(1) BW(2) Distributed/

Localized

Parameters

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.80

0.2

0.4

0.6

0.8

1

PA model -10 0 10

-90

-80

-70

-60

-50

Spectrum estimation

MethodMethodMethod

Check against spectral mask

ImpedanceVcc / Power Consumption

x

u

s

2argmin ux

2effectiveoutP

211

ISIS uxEVM

Projection of x onto u

122 su

Information Subcarriers

Effective B

W

Mu

ltiplexing

factor

Internal Gain

Fix allocation (approximately)Loop over methods, and parameters

Generate information signal (u)Generate TX signal (s)Loop over internal-gain

Calculate Pconsumption, Pout, EVMif spectral mask is violated

breakend if, end loopLoop over “channel” loss

Maximize SpectralEfficiency over internal-gainsEnd loop, End loopPlot SpectralEfficiency vs. loss

tionLossImplementa*

**

*

1

1

1

0

EVMNFWEffectiveBN

lossPSNRFinal

effectiveout

FinalSNRCP

WEffectiveBthput

1log*

1 2

OFDMBWthputngFactorMultiplexificiencySpectralEf /*

Performance approximation

13

“PAPR reduction” methods evaluation

17 18 19 20 21 22 23 24 25 2615

20

25

30

35

40

45

50

Effective Pout [dBm]

EV

M [

dB]

EVM comparison

OFDM, Max effective TX power=23.75 dBmClip&filt@TD, #iter=5, rejection=15dB, filt length=105, clip thres=4.0dB, Max effective TX power=25.14 dBm

Tone-resevation, Nreserved=140, clip thres=4.0dB, Max effective TX power=24.65 dBmTone-resevation w distortion, Nreserved=140, clip thres=4.0dB, Dstrt ratio=0.05, Max effective TX power=25.09 dBmTone-injection, offset=2.828427e+000, clip thres=2.5dB, clip gain=1.2, Max effective TX power=24.26 dBm

14

“PAPR reduction” methods evaluation

90 100 110 120 130 140 1500

1

2

3

4

5

6

7

Path loss [dB]

Sp

ect

ral e

ffic

ien

cy [

bits

/sec

/Hz]

Spectral efficiency comparison

OFDM

Clip&filt@TD, #iter=3, rejection=14dB, filt length=95, clip thres=5.5dB

Clip&filt@TD, #iter=3, rejection=17dB, filt length=95, clip thres=3.0dB

Tone-resevation, Nreserved=75, clip thres=5.5dB

Tone-resevation, Nreserved=150, clip thres=4.0dB

120 121 122 123 124 125 126 127 1280

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

90 100 110 120 130 140 1500

1

2

3

4

5

6

7

Path loss [dB]

Sp

ect

ral e

ffic

ien

cy [

bits

/sec

/Hz]

Spectral efficiency comparison

OFDM

Clip&filt@TD, #iter=3, rejection=14dB, filt length=95, clip thres=5.5dB

Clip&filt@TD, #iter=3, rejection=17dB, filt length=95, clip thres=3.0dB

Tone-resevation, Nreserved=75, clip thres=5.5dB

Tone-resevation, Nreserved=150, clip thres=4.0dB

120 121 122 123 124 125 126 127 1280

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Spectral efficiency vs. link loss

SE

[b/

s/hz

]

Link loss [dB]

15

Spectral efficiency of localized OFDMA

90 100 110 120 130 140 1500

1

2

3

4

5

6

7

8

9

10

Path loss [dB]

Sp

ec

tra

l e

ffic

ien

cy

[b

its

/se

c/H

z]

Spectral efficiency comparison

PUSC, #subch=2

PUSC, #subch=2

Clip&filt@TD PUSC, #iter=3, rejection=17dB, filt length=95, clip thres=3.0dB, #subch=2Narrrow band, Nsub-carr=48, Offset=0

Narrrow band, Nsub-carr=48, Offset=-1

134 136 138 140 142 144 146 1480

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

16

Localized gain as function of location in the band

-30 -20 -10 0 10 20 30-110

-100

-90

-80

-70

-60

-50

-40

-30

Frequency [MHz]

xx

(f)

[dB

m/H

z]

Spectral density and masks. TX power = 25.32

min spectrum

average spectrummax-hold spectrum

FCC mask

-30 -20 -10 0 10 20 30-110

-100

-90

-80

-70

-60

-50

-40

-30

-20

Frequency [MHz]

xx

(f)

[dB

m/H

z]

Spectral density and masks. TX power = 30.39

min spectrum

average spectrummax-hold spectrum

FCC mask

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 125

25.5

26

26.5

27

27.5

28

28.5

29

29.5

30Narrow band transmission, 48 tones

Normalized offset

Eff

ectiv

e T

X p

ower

[dB

m]

edgeedge edgeedge

centercenter

Signal bandwidthSignal bandwidth

17

Power limited PA• Our results are with PA peak

power limit (or fixed Vcc). • Another option is to consider

PA with a current limitation (modify Vcc to meet same power).

• In this case all differences in transmit powers are approximately halved (in dB)

TX

pow

er [d

Bm

]

Consumed power [dBm]

Maximum consumed power

(1)

(2)

TX power advantage for

constant power consumption

Maximum TX power without

“PAPR reduction” method:

limited by spectral mask and

power consumption

With “PAPR reduction” the

TX power can be increased,

but the consumed power

increases as well

Keeping constant backoff,

reducing Vcc and TX

power, there is advantage

in TX power without power

consumption penalty

(3)

A

B

C

1 Increase signal power while keeping constant Vcc, therefore reduce backoff

Slope

= 2

2 Scale signal power and Vcc together to keep constant backoff, that is keep constant proportion between signal and out-of-band emission. This can be associated with relative spectral mask.

Slope

= 1

3 Scale signal power while keeping the out of band emission power approximately constant.

Slope

= 0.8