SFH PHY Structure for IEEE 802.16m Amendment
Document Number: IEEE C802.16m-09/0977r1
Date Submitted: 2009-04-27
Source:
Pei-Kai Liao ([email protected]), Chih-Yuan Lin ([email protected]), Yih-Shen Chen, Paul Cheng
MediaTek Inc.
Venue:
Category: AWD comments / Area: Chapter 15.3.6 (DL-CTRL)
“Comments on AWD 15.3.6 DL-CTRL”
Base Contribution:
This is base contribution.
Purpose:
Propose to be discussed and adopted by TGm for IEEE 802.16m Amendment.
Notice:
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 >.
FR-1 + Interlaced Pilot Pattern +
Frequency-domain Repetitions



FR-1+ interlaced pilot pattern + frequency-domain
repetitions
Pilot power boosting , a = 5 dB
Claims:



Common structure to both DL data channel and control
channel so that same receiver design can be shared
Using MMSE-CNC receiver can achieve system
requirement under coding rate 1/24 (1/4 TBCC + 6
repetitions)
Capacity for SFH
 160 bits, if 24 PRUs are considered (24*80*2/24)
 133 bits, if 20 PRUs are considered (20*80*2/24)
Interlaced Pilot Patterns
Freq.
Time
data
pilot for stream 1
pilot for stream 2
Cell 1
Cell 2
Cell 3
MMSE-CNC Receiver

Received signal model:
N
y  H  s   Gi  di  n
i 1
White Noise
Color Noise
Signal Part

MMSE receiver

sˆ  Hˆ  Hˆ  Hˆ  Rˆ I
H
Rˆ I 
H
1
N PRU
N PRU
i 1

1
y
H
ˆ
ˆ
(
y

H

s
)

(
y

H

s
)
 i i
i
i
Interference Model Assumption: 6 Cells

Interfered environment



One serving BS
Five interfering BSs
 Two are first-tier and
three are
second-tier
Since FR = 1, BCH
messages
of three BS collide at
cell-edge MS
interfering BS 3
Pilot pattern 1
interfering BS 1
Pilot pattern 3
interfering BS 4
Pilot pattern 2
Serving BS
Pilot pattern 1
interfering BS 2
Pilot pattern 2
interfering BS 5
Pilot pattern 3
Analysis of Interference Covariance
Estimation (1/3)

Received signal model of data tones
5
y  H  s   Gi  di  n
i 1

Interference covariance of data tones
5
RI   Gi  Gi   n 2  ,
H
i 1
where it is assumed that the power of data tones is 1.
Analysis of Interference Covariance
Estimation (2/3)

Received signal model of pilot pattern 1
5
y  H  ps  G3  p3   Gi  d i  n
i 1,
i 3

Interference covariance of pilot pattern 1
5
RI  a  G3  G3   Gi  Gi   n 2  ,
H
H
i 1,
i 3
where a is the power boosting level of pilot tones.
Analysis of Interference Covariance
Estimation (3/3)



The interference covariance of data tone is different from that
of pilot tones due to pilot power boosting
Therefore, the interference covariance estimated by pilot tones
can not be used for data tones
It can only measure 2nd tier interferences, especially for cases
with large pilot power boosting
5
H
Rˆ I  a  G3  G3   Gi  Gi   n 2    E ,
H
i 1,
i 3
where a is the power boosting level of pilot tones,
E is the error matrix induced by channel estimation.
Remarks

Pros:



Common structure to both DL data channel and control channel so that
same receiver design can be shared
More data tones per PRU
Cons:




Different data mappings among different cells
Data-to-pilot collision induces large channel estimation error and more
bit errors in data channel so it may take more repetitions to achieve
system requirement
Higher pilot power boosting results in higher PAPR in time domain and
thus reduce the coverage of data tones
Interference covariance estimated by pilot tones is different from that of
data tones


It can only measure 2nd tier interferences, especially for cases with large pilot
power boosting
No optimal receiver for the transmission scheme of FR-1 + interlaced
pilot pattern + frequency-domain repetitions
MediaTek’s Proposal



FR-1 + interlaced pilot pattern + tone nulling + Frequency-domain
repetitions
Pilot power boosting, a = 3 or 5 dB [TBD]
Claims:




There is no data-to-pilot collision so that high accuracy of channel
estimation can be achieved even under low SIR
Lower pilot power boosting (ex. 3 dB) can be applied for reduced PAPR
and thus increase the coverage of data tones
Both 1/16 (1/2 CTC + 8 repetitions) and 1/12 (1/2 CTC + 6 repetitions)
can achieve similar performance as that of 1/24 using interlaced pilot
pattern only
Capacity for SFH





180 bits, if 24 PRUs are considered (24*60*2/16)
150 bits, if 20 PRUs are considered (20*60*2/16)
240 bits, if 24 PRUs are considered (24*60*2/12)
200 bits, if 20 PRUs are considered (20*60*2/12)
Flexibility for system upgrading due to higher SFH capacity
Proposed SFH PHY Structure for One PRU
Freq.
Time
null
data
pilot for stream 1
pilot for stream 2
Cell 1
Cell 2
Cell 3
Proposed MMSE Receiver (1/2)

Received signal model
N
y  H  s  G1  d1  G2  d 2   Gi  d i  n
i 3
White Noise
Interference Part
Signal Part

Proposed optimal MMSE Receiver

sˆ  Hˆ   Hˆ  Hˆ  Gˆ1  Gˆ1  Gˆ 2  Gˆ 2

H
Rˆ I ,s
1

a Np
H
H
Np
 ( y  Hˆ  s )  ( y  Hˆ  s )
i 1
Color Noise
i
i
i
i
i
i
H
H
1
3

ˆ
ˆ
ˆ
 RI , s  RI ,1  RI ,2  (  1)  ˆ n2     y
a

Rˆ I ,k
1

a Np
Np
 ( y  Gˆ
i 1
i
k ,i
 d k ,i )  ( yi  Gˆ k ,i  d k ,i ) H
Proposed MMSE Receiver (2/2)

Proposed suboptimal MMSE Receiver

sˆ  Hˆ  Hˆ  Hˆ  Gˆ1  Gˆ1  Gˆ 2  Gˆ 2  ˆ 
H

H
H
H
2
n

1
Only first-tier interferences are considered and other
interferences are considered as white noise or neglected
y
Interference Model Assumption: 6 Cells

Interfered environment


One serving BS
Five interfering BSs


Two are first-tier and
three are
second-tier
Since FR = 1, BCH
messages
of three BS collide at
cell-edge MS
interfering BS 3
Pilot pattern 1
interfering BS 1
Pilot pattern 3
interfering BS 4
Pilot pattern 2
Serving BS
Pilot pattern 1
interfering BS 2
Pilot pattern 2
interfering BS 5
Pilot pattern 3
Analysis of Interference Covariance
Estimation (1/5)

Received signal model of data tones
5
y  H  s   Gi  di  n
i 1

Interference covariance of data tones
5
RI   Gi  Gi   n 2  ,
H
i 1
where it is assumed that the power of data tones is 1.
Analysis of Interference Covariance
Estimation (2/5)

Received signal model of pilot pattern 1
y  H  ps  G3  p3  n

Interference covariance of pilot pattern 1
RI ,s  a  G3  G3   n 2  ,
H
where a is the power boosting level of pilot tones.
Analysis of Interference Covariance
Estimation (3/5)

Received signal model of pilot pattern 2
y  G2  p2  G4  p4  n

Interference covariance of pilot pattern 2
RI ,2  a  G4  G4   n 2  ,
H
where a is the power boosting level of pilot tones.
Analysis of Interference Covariance
Estimation (4/5)

Received signal model of pilot pattern 3
y  G1  p1  G5  p5  n

Interference covariance of pilot pattern 3
RI ,1  a  G5  G5   n 2  ,
H
where a is the power boosting level of pilot tones.
Analysis of Interference Covariance
Estimation (5/5)


The covariance matrix of each single pilot pattern is
different from that of data tones
However, the sum of these covariance matrices can
be used for covariance matrix calculation for data
tones


H
H
1
3 
ˆ
ˆ
ˆ
ˆ
ˆ
RI  G1  G1  G2  G2   Rˆ I , s  Rˆ I ,1  Rˆ I ,2    1  ˆ n2    E
a
a 
5
H
 Gˆ1  Gˆ1  Gˆ 2  Gˆ 2   Gi  Gi  ˆ n 2    E ,
H
H
i 3
where a is the power boosting level of pilot tones,
E is the error matrix induced by channel estimation.
Issues of Proposed Receiver Scheme

If PRBS sequence is applied to pilot tones, how does
AMS know the PRBS sequences of the other two
pilot patterns?



SFH is decoded immediately after synchronization
This information can be obtained from SA-Preamble by
knowing cell ID if the PRBS sequence assignment depends
cell ID
Since three interlaced pilot patterns align with three
segments of SA-Preamble, there is no problem for an AMS
to obtain this information by SA-Preamble
Remarks

Pros:







Same data mappings among all cells
There is no data-to-pilot collision so that accurate channel estimation can
be achieved and the bit errors in data channel can be reduced so it may
take less repetitions to achieve system requirement
Flexibility to system upgrade by reserving more SFH bits
Lower pilot power boosting can be applied and thus increase the
coverage of data tones
Accurate interference covariance estimation can be obtained
There is an optimal receiver for the transmission scheme of FR-1 +
interlaced pilot pattern + tone nulling + frequency-domain repetitions
Cons:


No common structure to both DL data channel and control channel so
that the receiver design for SFH may be different from data channel
Less data tones per PRU
Simulation Parameters





2x2 MIMO SFBC system with 512-size FFT
Modulation/coding: QPSK ½ + 6, 8 or 12 repetitions
Optimal whole-band MMSE-based combing
Channel model: VA 120
2D MMSE channel estimator

PRU-based channel estimation





2-PRU CE window for interlaced pilot pattern
1-PRU CE window for interlaced pilot pattern + tone nulling
Pilot power boosting: 5 dB
Noise level: INR = 10 dB
Interference limited environment

3-cell case: interference ratio: 0.5:0.5


1 serving BS, two interferers
6-cell case: interference power ratio: 0.35 : 0.35 : 0.1 : 0.1 : 0.1

1 serving BS, five interferers
Interference Scenarios

Interfered environment

3-cell case





One serving BS
Two interfering BSs
Serving BS
6-cell case


interfering BS 1
One serving BS
Five interfering BSs
Since FR = 1, SFH
messages
of three BS collide at
cell-edge MS
All pilot values are
assumed to be 1 x power
boosting level, no PRBS
sequence applied
interfering BS 2
Cell-edge MS
interfering BS 3
Pilot pattern 1
interfering BS 1
Pilot pattern 3
interfering BS 4
Pilot pattern 2
Serving BS
Pilot pattern 1
interfering BS 2
Pilot pattern 2
interfering BS 5
Pilot pattern 3
Receiver Scheme for Simulation

Sub-optimal MMSE-CNC Receiver

sˆ  Hˆ  Hˆ  Hˆ  Gˆ1  Gˆ1  Gˆ 2  Gˆ 2  ˆ 
H

H
H
H
2
n

1
Only first-tier interferences are considered and other
interferences are considered as white noise or neglected
y
Simulation Results (1/2)
3-cell case: VA 120 channel model (INR = 10 dB)
For interlaced only structure, 2-PRU CE window is used, but only 1-PRU CE
window is used for interlaced nulltone structure

Interlaced nulltone 8 rep.
Interlaced nulltone 6 rep.
Interlaced 8 rep.
Interlaced 12 rep.
-1
10
PER

-2
10
-12
-11
-10
-9
-8
SINR (dB)
-7
-6
-5
-4
Simulation Results (2/2)
6-cell case: VA 120 channel model (INR = 10 dB)
For interlaced only structure, 2-PRU CE window is used, but only 1-PRU CE window is
used for interlaced nulltone structure

0
10
Interlaced nulltone 8 rep. (perfect)
Interlaced nulltone 8 rep.
Interlaced nulltone 6 rep. (perfect)
Interlaced nulltone 6 rep.
Interlaced 12 rep. (perfect)
Interlaced 12 rep.
PER

-1
10
-2
10
-9
-8
-7
-6
-5
-4
SINR (dB)
-3
-2
-1
0
1
Conclusion





For FR-1, there is no optimal receiver scheme if only
interlaced pilot pattern in current AWD is applied
With tone nulling in FR-1, optimal receiver is available
Even with suboptimal receiver, MediaTek’s design can easily
achieve the same performance using code rate 1/16 as that of
1/24 so as to provide capacity of 180 bits at most, which is
larger than 160 bits provided by interlaced pilot pattern only
If A-MAP should also be transmitted in the 5Mhz, SFH
capacity issue becomes more critical and thus requires higher
code rate and MediaTek’s design still can provide capacity of
150 bits for 20 PRUs
It is recommended to adopt MediaTek’s proposal for FR-1 SFH
PHY structure design
Text Proposal (1/3)
[Add the following text into the TGm AWD]
-------------------------------Start of the Text----------------------------------------------------15.3.6.2.1 Superframe Header
…………………………
The PHY structure of a PRU for resource allocation of the SFH is described in Figure
X Section <<15.3.5>>. The SFH is transmitted within a predefined frequency partition
called the SFH frequency partition. The SFH frequency partition consists of NPRU, SFH
PRUs within a 5 MHz physical bandwidth.
The PRUs in the SFH frequency partition uses the 2-stream pilot pattern defined in
<<15.3.5>>. The PRUs in the SFH frequency partition are permuted to generate
NPRU, SFH distributed LRUs.
Text Proposal (2/3)
6 Symbols
1
6 Symbols
1
1
1
1
Pilot Pattern Set 0
1
1
1
1
Pilot Pattern Set 1
(a) Stream one
1
18 Contiguous Subcarriers
18 Contiguous Subcarriers
18 Contiguous Subcarriers
1
6 Symbols
1
1
1
1
Pilot Pattern Set 2
Text Proposal (3/3)
2
2
Pilot Pattern Set 0
2
2
2
Pilot Pattern Set 1
18 Contiguous Subcarriers
18 Contiguous Subcarriers
18 Contiguous Subcarriers
2
2
2
2
2
2
6 Symbols
6 Symbols
6 Symbols
2
2
2
2
Pilot Pattern Set 2
(b) Stream Two
Figure X — PHY Structure of a PRU in SFH
-------------------------------End of the Text-----------------------------------------------------