Nov. 2013
doc.: IEEE 802.11-13/1391r0
About SINR conversion for PHY
Abstraction
Date: 2013-11-11
Authors:
Name
Affiliations
Address
Phone
email
Yakun Sun
Marvell
Semiconductor
5488 Marvell Ln, Santa Clara,
CA 95054
1-408-2223847
[email protected]
Yan Zhang
Marvell
Semiconductor
Hongyuan Zhang
Marvell
Semiconductor
Hui-Ling Lou
Marvell
Semiconductor
Mingguang Xu
Marvell
Semiconductor
Submission
Slide 1
Yakun Sun, et. Al.
Nov. 2013
doc.: IEEE 802.11-13/1391r0
Introduction
• Effective SINR mapping is proposed for PHY
abstraction [1, 2, 3].
• The effective SINR mapping, no matter what specific
mapping function is adopted, operates in the same
procedure.
– Starting by calculating receiver-output SINR for each tone.
• We discuss how to calculate receiver-output SINR.
Submission
Slide 2
Yakun Sun, et. Al.
Nov. 2013
doc.: IEEE 802.11-13/1391r0
PHY Layer Model for System Simulation
•
•
•
MAC Layer
Tx MAC layer informs PHY layer the
number of bits and MCS.
Tx PHY layer does not encode
anything  a virtual PHY packet.
PHY params: number
of bits, MCS, etc.
Channel includes (1) large scale
fading, and (2) instantaneous channel
impulse response (CIR)
–
Virtual Encoder/
Modulator
Generated for performance
evaluation.
Multiple virtual PHY layers may
contribute to the received “signals”.
–
•
•
•
•
No encoding/
modulation is
actually done
Virtual Decoder
{SINR(n,l)}SNR_effPER
Virtual MIMO
(CSD/STBC/SM/TxBF)
{SINR(n,l), n=1...N_tones,
l=1...N_OFDM_symbols}
Virtual Receiver
(FFT, MIMO detector)
Precoders
Each with a CIR.
Rx calculates channel freq response.
Rx MIMO detector only calculates
SINR but does not process any
signals.
Rx decoder takes SINRs across
frequency tones and OFDM symbols,
and predict PER for this packet.
A random number is generated and
compare with PER
Decoded correctly or
not? (draw a random
number and compare
with predicted PER)
Number of OFDM symbols
PHY
•
MAC Layer
Transmitter Frontend
Abstraction
TxEVM, Tx power,
antenna gain, etc
Thermal noise,
noise figure,
antenna gain, etc.
Receiver Frontend
Abstraction
Propagation
Channel
Propagation
Channel
Interfering Transmitter
PHY layer
Submission
Slide 3
Yakun Sun, et. Al.
Nov. 2013
doc.: IEEE 802.11-13/1391r0
Frequency Domain Received Power
• Frequency domain equalization is done and the PER
depends on the equalizer performance.
• Frequency domain received signal power is calculated on
top of the channel frequency response.
– A scaling factor to compensate the guard tones.
– A simple example (more factors can be added, such as cable loss…):
P  dBm   PTX  dBm   GTX  dBi   PL  dB   SF  dB   GRX  dB   Scaling  dB 
PTX : transmit power
GTX / RX : transmit/receive antenna gain
PL : path loss
SF : shadowing factor
 N FFT
Scaling  10 log10 
N
 used _ tones
Submission



Slide 4
Yakun Sun, et. Al.
Nov. 2013
doc.: IEEE 802.11-13/1391r0
Frequency Domain Received Signal Model
• The (virtually) received signal at the nth tone, lth OFDM symbols is
K
r  n, l   P0 H 0  n, l  W0 x0  n, l    Pj H j  n, l  W j x j  n, l   n  n, l 
Desired transmitter
j 1
Interference
n 1
N data _ tones , l  1
NOFDM _ sym
H j : Instantaneous MIMO channel frequency response for the j th transmitter.
W j : Precoding matrix per tone for the j th transmitter
Pj : Received signal power in frequency domain for the j th transmitter
x j : (Virtually) transmitted symbol vector from the j th transmitter.
n : receiver noise
• Instantaneous channel fading is separated from the received signal
power.
– The power of channel frequency response |H(f)|2 is normalized, either per
packet or long-term normalized.
• The received noise is modeled as AWGN with variance σ2.
 2  dBm  N0  dBm / Hz   BW  NF
Submission
Slide 5
Yakun Sun, et. Al.
Nov. 2013
doc.: IEEE 802.11-13/1391r0
SINR Calculation
•
Based on the receiver assumed, the equalizer output SINR can be
calculated for each spatial stream.
•
SINR is precoding (beamforming) dependent.
– Per-tone beamforming changes the effective channel fading rather than a
constant/static receive signal power boost.
– The beamforming method has to be aligned or specified in simulation.
•
SINR is receiver dependent.
– Typically, linear MMSE (with or without interference whitening) or MRC for
single-stream transmission is assumed.
– “SINR” does not exist for ML detector, and equivalent SINR for ML has been
proposed in literatures but with great computational complexity.
•
For example, if no interference is present (K=0), and a rank-1
transmission (x0 is a scalar). The MRC output SINR is:
r  n, l   P0 H 0  n, l  W0 x0  n, l   n  n, l   SINR  n, l  
Submission
Slide 6
P0 H 0 W0
2
2
Yakun Sun, et. Al.
Nov. 2013
doc.: IEEE 802.11-13/1391r0
Ideal vs. Practical Channel Estimation
• Receiver can assume ideal channel estimation.
– Textbook SINR equations
• If practical channel estimation is assumed, the impact
on SINR can be represented by an additional noise for
channel estimation error.
– For example, a good approximation for SISO channel with oneshot channel estimation is to scale noise power by 2, (i.e., σCE2=
2σ2)
– Better approximations for CE errors are open to discuss.
Submission
Slide 7
Yakun Sun, et. Al.
Nov. 2013
doc.: IEEE 802.11-13/1391r0
Comments on SINR Calculation
• Given asynchronous transmission in OBSS, the transmit
power and channel responses from each interfering BSS
may vary (and disappear) across time.
• K interfering transmitters are assumed to be frequencyselective.
– To simplify the simulation, flat-fading channels can be assumed for
transmitters far away (with low received power).
• TxEVM can be modeled by capping EVM-free SINR.
SINRNoEVM  n, l 
SINREVM  n, l  
SINRNoEVM  n, l  EVM  1
Submission
Slide 8
Yakun Sun, et. Al.
Nov. 2013
doc.: IEEE 802.11-13/1391r0
Summary
• Briefly introduce how the receiver-output SINR can be
calculated.
• Receiver-output SINR is a function of channel
frequency responses and received signal power from
each transmitter (desired or not).
• SINR depends on the receiver type and beamforming
schemes.
• SINR calculation can include a simple and efficient
modeling of transmitter/receiver details, such as CE
error, EVM, and etc.
Submission
Slide 9
Yakun Sun, et. Al.
Sept 2013
doc.: IEEE 802.11-13/1391r0
References
[1] IEEE 802.16m-08/004r5, Jan. 2009
[2] R1-050680, “Text Proposal: Simulation Assumptions and Evaluation for
EUTRA”, 3GPP TSG RAN WG1 #41bis, June, 2005
[3] R1-061626, “LTE Downlink System Performance Evaluation Results”, 3GPP
TSG RAN1 #45, May, 2006
Submission
Slide 10
Yakun Sun, et. Al.