1. Partial Feedback Scheme with an Interference‐
Aware Subcarrier Allocation Scheme in a
Correlated LTE Downlink
Rosdiadee Nordin, Mahamod Ismail
Department of Electrical, Electronics and Systems Engineering
Faculty of Engineering and Built Environment
Universiti Kebangsaan Malaysia (Malaysia)
adee@eng.ukm.my, mahamod@eng.ukm.my
INTRODUCTION
• Full feedback scheme increase the uplink
overhead requirement
• This paper consider the use of partial
feedback scheme utilizing DFT‐based
codebook precoding to exploit spatial
diversity from the multiuser (MU)‐MIMO.
• In addition, the frequency diversity will be
exploited via an interference‐aware
subcarrier allocation scheme.
• Self‐interference occurs due to an increasing
spatial correlation between the
communicating MIMO antennas
PROBLEM
BACKGROUNDS
Feedback Schemes in MU‐MIMO Transmissions
• MU‐MIMO transmission uses the Channel Quality
Information (CQI) to serve the spatially
multiplexed users from the precoding techniques
• The use of precoding is at the expense of channel
knowledge – places significant burden on the
uplink
• Two feedback strategies considered:
– Full: feeds back a single CQI value for every matrix in
the codebook for each RB
– Partial: feeds back a single CQI value for the preferred
matrix for each RB (quantized)
• CQI information available per RB, allows channel
resources to be allocated effectively to different
users while reducing the amount of feedback.
DFT‐Based Codebook Precoding
• In LTE Rel 8, eNodeB transmits through a
codebook‐based spatial beam, which ensures
uniform sector coverage across the cell
• DFT‐based codebook has shown to be effective
against wide range of spatial correlation:
uncorrelated to fully correlated
-10 -5 0 5 10 15 20
0
2
4
6
8
10
12
SNR (dB)
capacity(bps/Hz)
RBS
=0.0,RMS
=0.0
RBS
=0.4,RMS
=0.4
RBS
=0.5,RMS
=0.5
RBS
=0.0,RMS
=0.9
RBS
=0.9,RMS
=0.0
RBS
=0.9,RMS
=0.9
RBS
=1.0,RMS
=1.0
OBJECTIVES
• Investigate BER performance between
partial vs. full feedback scheme in varying
MIMO channel conditions
• Mitigate the effect of MIMO spatial
correlation (self‐interference) by combining
both spatial and frequency diversity

2. METHODOLOGY
• SINR metric, representation of CQI & to determine the subcarrier allocation process
• has the knowledge of self‐interference, especially when the correlation is high inside spatial
subchannels
• Allows the user with lowest channel
gain to have the next best subcarrier
gain: fairness vs. error probability
• Involves sorting, comparing and
simple arithmetic.
• Ranks users from lowest to highest
channel gain
 
  NGGEHG
EHG
SINR
qjqjkqqksqjqjkk
sqqkkq
k




 


2
,
22
,
2
Knowledge of self-
interference
Main spatial layerq= spatial layer
k= subcarrier index
MMSE filter
Parameters Value
Downlink Bandwidth 5 MHz
Time Slot/ Sub‐frame
duration
0.5 ms/
1ms
Subcarrier Spacing 15 kHz
Precoding CB size, L 1,2,4,8
FFT Size, NFFT 1024
Useable subcarrier, Nsub 600
Total users, K 10
OFDM symbols/ time slot
(Short/Long CP)
7/6
Correlation Modes RMS RBS
Uncorrelated 0.00 0.00
Fully Correlated 0.90 0.90
Feedback Scheme
Full
Feedback
Partial
Feedback
SU‐
MIMO
Preferred Layer 1 CQI 4 bits 4 bits 4 bits
Preferred Layer 2 CQI 4 bits 4 bits
Alternative Layer 1 CQI 4 bits ‐ ‐
Alternative Layer 2 CQI 4 bits ‐ ‐
Preferred Matrix Index 1 bit 1 bit 1 bit
Total bits per RB 17 bits 9 bits 5 bits
Assumptions
• 2x2 MU‐MIMO, QPSK ½ rate (LTE
Rel. 8)
• 500m radius, NLOS with 251 ns
delay spread (3GPP‐SCM Urban
Micro)

3. RESULTS & ANALYSIS
-8 -6 -4 -2 0 2 4 6 8
10
-3
10
-2
10
-1
10
0
Signal-to-Noise Ratio (SNR) in dB
BitErrorRate(BER)
CSI only (no precoding)
DFT only, L=2
DFT+ Interference-Aware, L=2
DFT+ Interference-Aware, L=8
Fully correlated channel Uncorrelated channel
-8 -6 -4 -2 0 2 4 6 8
10
-3
10
-2
10
-1
10
0
Signal-to-Noise Ratio (SNR) in dB
BitErrorRate(BER)
Partial MU, RMIMO
=0.99
Full MU, RMIMO
=0.99
Partial MU, RMIMO
=0.00
Full MU, RMIMO
=0.00
CONCLUSIONS
• DFT‐based codebook adaptation enables the quantization to exploit the spatial correlation
inherent in the channel.
• Full feedback scheme offers superior BER performance at the expense of a high uplink
overhead requirement.
• The partial feedback scheme offers a trade‐off between the multiuser diversity gain and
reduced feedback requirement.
• Combination of partial feedback and interference‐aware subcarrier allocation scheme improve
the BER performance, especially in a fully correlated MIMO channel.
-10 -5 0 5 10 15 20
10
-3
10
-2
10
-1
10
0
Signal-to-Noise Ratio (SNR) in dB
BitErrorRate(BER)
L=1, 'Full'
L=1, 'Uncorr'
L=2, 'Full'
L=2, 'Uncorr'
L=4, 'Full'
L=4, 'Uncorr'
L=8, 'Full'
L=8, 'Uncorr'
• SU‐MIMO does not benefit from the increased
codebook size for either correlation scenario since
eNodeB unable to exploit the channel knowledge from
the feedback path
• partial feedback scheme offers negligible
performance loss to the full feedback scheme with
the advantage of a reduced overhead requirement
on the uplink capacity
• Benefit of combining the DFT‐based codebook
precoding and an interference‐aware subcarrier
allocation scheme in a fully correlated channel for
partial feedback MU‐MIMO case
-8 -6 -4 -2 0 2 4 6 8
10
-3
10
-2
10
-1
10
0
Signal-to-Noise Ratio (SNR) in dB
BitErrorRate(BER)
CSI only (no precoding)
DFT only, L=2
DFT+ Interference-Aware, L=2
DFT+ Interference-Aware, L=8
• Trade‐offs to mitigate self‐interference vs.
codebook size
• Larger codebook has a richer selection of
precoding matrices that can be used to find a
better codeword match during encoding
*This work is supported by the Universiti Kebangsaan Malaysia, under the grant scheme UKM‐GGPM‐ICT‐032‐2011*