Codebook Design for LTE-A Downlink System

Lu Wu, Jinhui Chen, Hongwei Yang and Di Lu

Research & Innovation Center, Alcatel-Lucent Shanghai Bell, Shanghai, China
Email: {lu.wu, jinhui.chen, hongwei.yang, di.a.lv}@alcatel-sbell.com.cn

Abstract—Multiple-input multiple-output (MIMO) has been which selects a cluster of beams first, then performs beam
adopted by long term evolution (LTE) and its updated version selection and co-phasing between two antenna subgroups. But
LTE-Advanced (LTE-A) to increase the spectrum efficiency. this codebook design does not perform well in co-polarized
However, the benefits of multi-user (MU) MIMO highly rely on
the accurate channel knowledge at the transmitter. So proper linear antenna array since it does not fully match the channel
codebook design is a key problem for LTE-A FDD downlink characteristic of this antenna configuration due to the limited
system. In this paper, a codebook design with phase adjustment in co-phasing.
the precoder targeting wideband channel properties is proposed. In this paper, we indicate that precoder should have the
The average cell spectrum efficiency and cell-edge user spectrum form of DFT vector for closely-spaced co-polarized linear
efficiency of the proposed scheme are evaluated through system
level simulation and compared to the discrete Fourier transform antenna array due to the high spatial correlation. Then an
(DFT) based feedback scheme. The results show that the proposed enhanced DFT based codebook is proposed herein with phase
codebooks significantly improve feedback efficiency and system adjustment added in the precoder targeting wideband CSI. By
performance for co-polarized linear antenna array. use of this phase adjustment, the channel characteristic of co-
polarized arry can be fully utilized as much as possible without
I. I NTRODUCTION
increasing codebook size. Hence, compared to DFT based
Multiple-input multiple-output (MIMO) spatial multiplex- feedback [6], CSI feedback efficiency and system performance
ing is recognized to be capable of dramatically increasing with the proposed codebook can be improved significantly
the spectrum efficiency of wireless communication [1] and for co-polarized antennas. As to cross-polarized linear antenna
has been adopted by the 3rd Generation Partnership Project array, since the feedback granuality is the same, the proposed
(3GPP) long term evolution (LTE) and its updated version codebook can still perform well as DFT based feedback, which
LTE-Advanced (LTE-A) as one of the key techniques. High is also demonstrated by system level simulations.
data rates (326 Mbps for LTE [2], 1 Gbps for LTE-A [3][4]) The remainder of this paper is organized as follows. Section
and high spectrum efficiency are some of the requirements for II introduces the downlink system model. Section III firstly
these standards. These ambitious targets can be achieved only describes DFT based feedback, then derives the proposed
by the usage of advanced MIMO techniques. codebook. System level simulation results are presented in
It is well known that multi-user (MU) MIMO is very sen- Section IV. Finally, Section V provides some concluding
sitive to the accuracy of channel knowledge at the transmitter. remarks.
Contrary to single user (SU) MIMO, limited feedback in Notation: (·)T is transpose, nmodN denotes the modulus
MU-MIMO limits the gain of spatial multiplexing due to function, and diag(a1 , a2 , · · · , aN ) represents a diagonal ma-
the quantization error. In other words, as signal-to-noise ratio trix with a1 , a2 , · · · , aN being the main diagonal elements.
(SNR) increases, MU-MIMO becomes interference limited
due to the intra-cell interference induced by the quantization II. T HE D OWNLINK S YSTEM M ODEL
error. In 3GPP LTE-A, MIMO and orthogonal frequency division
In a frequency division duplex (FDD) system, the transmit- multiplexing (OFDM) are employed for downlink system.
ter gets the channel state information (CSI) via quantization Suppose that each base station (BS) has Nt transmit antennas
and limited feedback from each user. In recent 3GPP meetings, and each user is equipped with Nr receive antennas. At BS,
it has been agreed that the feedback framework allows the the transmitted signal for subcarrier k can be expressed as
report of two matrices selected from two separate codebooks x(k) = F(k)s(k) (1)
[5]. One matrix targets wideband and/or long-term channel
T
properties, and the other matrix targets frequency-selective where s(k) = [s1 (k), s2 (k), · · · , sR (k)] is a R × 1 vector
and/or short-term channel properties. Hence, the design for containing the encoded MIMO complex data symbols at
the two codebooks is a main problem. In practice, co-polarized subcarrier k, R is the rank of the transmission, and F(k) is a
linear antenna array is a typical configuration in LTE-A sys- Nt × R complex precoding matrix.
tems. There should be a proper codebook design corresponding Codebook based precoding is employed for LTE-A FDD
to this antenna array. In LTE-A discussions, many feedback system. Each user selects the codeword in the predefined
proposals have been presented to quantize the two matrices. codebook to represent the CSI from the served BS, and
Among these schemes, discrete Fourier transform (DFT) based sends the preferred codeword index to the BS. There are
feedback [6] is a smart design and receives the most concern, some criterias to select codeword, such as maximum capacity

978-1-4244-8327-3/11/$26.00 ©2011 IEEE

and chordal distance. Then BS retrieves the downlink CSI where ek denotes a Nb × 1 elementary vector with all zeros
according to the codeword index fed back from user, and except for the (k + 1)-th element with value 1. For rank 2,
performs proportional fairness scheduling by greedy search  
1 Y1 Y2
based on maximum weighted sum capacity. If SU mode is W2 = √ (6)
2 bY1 −bY2
scheduled, the precoding matrix F(k) is set as the codeword
fed back by the corresponding user. Otherwise, if MU mode where b represents co-phasing, and Y1 , Y2 are selected from
is scheduled, F(k) is usually calculated by zero-forcing (ZF) the elementary vectors.
method based on the downlink CSI of each scheduled user. With proper selection of a and b, the design in equations
(3)-(6) can also be made to suit the closely spaced co-polarized
III. C ODEBOOK D ESIGN linear array.
How to design the codebook effectively quantizing the Assume that the whole band is divided into M subbands.
CSI is a very important question for FDD downlink system. The sizes of C1 and C2 are denoted by N1 and N2 bits
In recent 3GPP meetings, it has been agreed that the CSI respectively. At first, user selects codeword W1 in codebook
feedback framework allows the report of two matrices selected C1 on the whole band. Then on each subband, user selects
from two separate codebooks. One matrix targets wideband codeword W2 in codebook C2 to match the subband CSI.
and/or long-term channel properties, and the other matrix tar- Hence, the overhead of DFT based feedback is N1 + M N2
gets frequency-selective and/or short-term channel properties. bits.
Hence, the two codebooks need to be designed properly. Based on the design described above, 8-Tx codebooks for
DFT based feedback are concretely presented in [6] for co-
A. DFT Based Feedback polarized and cross-polarized antennas. 128 beams are gener-
In recent LTE-A meetings, a two-stage codebook design is ated from the multiplication of W1 and W2 , in which only
agreed in [6]. The recommended codeword W for a subband sixteen beams have the form of 8-Tx DFT vectors matching
is the spatial characteristic of co-polarized antenna setup. In W1 ,
W = W1 W2 (2) X is composed of four adjacent overlapping beams. Each
beam is one of 32 4-Tx DFT vectors. The use of adjacent
where W is a Nt × R matrix, the outer matrix W1 ∈ C1 is a
overlapping beams is to reduce edge effect in frequency-
Nt × Nb matrix representing the wideband CSI, and the inner
selective precoding. Thus sixteen W1 codewords could be
matrix W2 ∈ C2 is a Nb × R matrix.
obtained as
This structure is well-suited for efficiently supporting com-  
(0) (1) (15)
mon antenna setups such as closely spaced cross-polarized or C1 = W1 , W1 , · · · , W1
co-polarized linear array. To see how correlations properties  (k) 
(k) X 0
are exploited, first consider the common case of an array of W1 = , k = 0, 1, · · · , 15 (7)
0 X(k)
closely spaced cross-polarized antennas. The antennas can then
be divided into two separate subgroups depending on the polar- where
 
ization direction of the antenna. The correlation is high among X(k) = b2kmod32 b(2k+1)mod32 · · · b(2k+3)mod32
the channels within an antenna subgroup while channels from  2πm 2πm∗2

2πm∗3 T
different antenna subgroups fade in an independent manner, bm = 1, ej 32 , ej 32 , ej 32 , m = 0, 1, · · · , 31.
and to some extent with reduced cross-talk due to the use of (8)
orthogonal polarizations. Such an antenna setup thus creates
W2 codebook C2 is composed of column selection and QPSK
quite pronounced channel properties, which are well-matched
co-phasing, and is expressed as follows:
to a block diagonal structure of W1 , i.e.,
  • Rank 1
X 0
W1 = (3)
   
0 X 1 Y 1 Y
W2 ∈ C2 = √ ,√ ,
where X targets a co-polarized antenna subgroup. Since the 2  Y  2 jY 
correlation is high within the antenna subgroup, it makes sense 1 Y 1 Y
to use a grid of beam codebook implemented from DFT based √ ,√
2 −Y 2 −jY
precoder vectors. Thus, X is composed by Nb Nt × 1 adjacent
Y = {e0 , e1 , e2 , e3 } (9)
DFT vectors.
The inner matrix W2 selects one beam in W1 and adjusts • Rank 2

 
the relative phase shift between polarizations. For rank 1, W2 1 Y 1 Y2
W2 ∈ C2 = √ ,
could be formed as
  2  Y1 −Y2 
1 Y 1 Y1 Y2
W2 = √ (4) √
2 aY 2 jY1 −jY2
In the equation above, a represents co-phasing, and (Y1 , Y2 ) ∈ {(e0 , e0 ), (e1 , e1 ), (e2 , e2 ), (e3 , e3 ),
Y ∈ {e0 , e1 , · · · , eNb −1 } (5) (e0 , e1 ), (e1 , e2 ), (e0 , e3 ), (e1 , e3 )} (10)
B. Proposed Codebook If an unified codebook is needed for both co-polarized and
For closely-spaced co-polarized linear array, the dominated cross-polarized antenna, the codebook proposed above can be
eigen-vector of the channel correlation matrix can be approxi- modified by just adding a co-phasing factor a in W2 , which
mated by a DFT vector due to the high spatial correlation. targets the phase adjustment between two antenna subgroups
So the rank-1 precoder W needs to be designed as DFT for cross-polarized antenna. Then W2 is constructed as
vectors specific to this antenna configuration. In the DFT • Rank 1
 
based feedback described in Section III-A, the co-phasing in 1 Y
W2 = √
W2 is important for matching the phase shift between the 2 aY
two antenna subgroups. However, due to the limited selection Y = {e0 , e1 , · · · , eNb −1 } (17)
of co-phasing, 8-Tx codebooks for DFT based feedback only
provide sixteen DFT vectors from total 128 beams. If more • Rank 2
 
DFT vectors are wanted, it needs more bits for co-phasing, 1 Y Y
W2 = √
which must increase the codebook size and the feedback 2 aY −aY
overhead. In this section, an enhanced two-stage codebook Y = {e0 , e1 , · · · , eNb −1 } (18)
is proposed to overcome this problem without increasing the
overhead of DFT based feedback. Next, as an example, the proposed 8-Tx codebook is pre-
Define the N × 1 DFT vector as sented in the following.
  
2πm(N −1) T (0) (1) (15)
m
fM,N
2πm
= 1, ej M , · · · , ej M , m = 0, 1, · · · , M − 1 C1 = W1 , W1 , · · · , W1
 (k) 
(11) (k) X 0
W1 = , k = 0, 1, · · · , 15 (19)
where M/N denotes the value of oversampling. It is obvious 0 X(k) Λ(k)
that the phase shift between two subgroups is fixed for the
where
(m + 1)-th DFT vector, i.e., 2πm(N/2)/M . Basing on this  
observation, we indicate that for co-polarized antenna, there is X(k) = b2kmod32 b(2k+1)mod32 · · · b(2k+3)mod32
 2π4
no need to feed back any bit for co-phasing due to the fixed 2π4
Λ(k) = diag ej 32 (2kmod32) , ej 32 ((2k+1)mod32) ,
phase shift of each beam. In the following, we propose adding 
the fixed phase shift of each beam into W1 in equation (3) 2π4 2π4
ej 32 ((2k+2)mod32) , ej 32 ((2k+3)mod32)
to construct DFT vectors. Furthermore, an unified codebook 
2πm 2πm∗2 2πm∗3 T
design is provided for both co-polarized and cross-polarized bm = 1, ej 32 , ej 32 , ej 32 , m = 0, 1, · · · , 31.
linear array. (20)
Define Z(k) as follows, which is comprised by Nb DFT
vectors. And W2 codebook is written as
 kNb
• Rank 1
k1 k2
Z(k) = fM,N , f , · · · , f (12)  
t /2 M,Nt /2 M,Nt /2 1 Y
W2 = √
Define
 2 aY
2πk1 (Nt /2) 2πk2 (Nt /2) 2πkN (Nt /2)
b Y = {e0 , e1 , e2 , e3 }
Λ(k) = diag ej M , ej M , · · · , ej M

{1}, co-polarized antenna

(13) a∈ (21)
{1, −1, j, −j}, unified codebook
Then we propose W1 as follows:
 (k) • Rank 2
(k) Z 0  
W1 = , k = 0, 1, · · · , N1 − 1 (14) 1 Y Y
0 Z(k) Λ(k) W2 = √
2 aY −aY
With the diagonal matrix Λ(k) added, each corresponding Y = {e0 , e1 , e2 , e3 }
columns in the first and second block diagonal matrix can

{1}, co-polarized antenna

construct Nt -Tx DFT vectors. Thus, for co-polarized antenna, a∈ (22)
{1, −1, j, −j}, unified codebook
only beam selection is needed in W2 , i.e.,
• Rank 1
Obviously, compared to DFT based feedback in equations
  (7)-(10), the proposed codebooks presented above reduce
1 Y
W2 = √ payload size of 2M bits for co-polarized antenna. Besides,
2 Y since the proposed codebooks provide the largest number of
Y = {e0 , e1 , · · · , eNb −1 } (15) DFT vectors without increasing codebook size, the system
• Rank 2 performance can be further improved as the reduction of
  beam granularity. As to cross-polarized linear antenna array,
1 Y Y
W2 = √ although the proposed codebooks use different co-phasing to
2 Y −Y DFT based feedback, there is no big difference between their
Y = {e0 , e1 , · · · , eNb −1 } (16) performance due to the same co-phasing granularity.
 TABLE I
S IMULATION PARAMETERS FOR LTE-A D OWNLINK 1

Deployment scenario 3GPP case 1 3D, SCM-UMa, 0.9

8 degree angle spread
0.8
Cell number 19 cells with 3 sectors per cell
Wrap-around model Yes 0.7
Duplex method and bandwidth FDD: 10MHz for downlink
Network synchronization Synchronized 0.6
Traffic model Full buffer

CDF
0.5
User number per sector 10
Maximal number of 0.4
co-scheduled users 8 for 8Tx
BS Antenna 8Tx: Co-polarized/Cross-polarized 0.3
antennas with the spacing of half
wave length, polarization angles 0.2
of 90◦ /±45◦
0.1 DFT based codebook
User antenna 2Rx: half wave length spacing; Proposed codebook
co-polarized antennas with vertical 0
polarization; cross-polarized 0 2000 4000 6000 8000 10000 12000 14000
User Throughput (kbps)
antennas with polarization angles
of 90◦ /0◦
Transmission scheme ZF method, MU-MIMO, Fig. 1. CDF of user throughput for 8Tx co-polarized antenna.
rank 1 per user
Scheduler Proportional fair and frequency
selective scheduling, scheduling
granularity of one subframe (1 ms) 1
Feedback Wideband W1 , subband W2 and
subband CQI: 5ms report 0.9
periodicity, 6ms feedback delay
0.8
Link adaptation Error: N(0,1dB) per PRB
HARQ Synchronous HARQ, chase 0.7
combining, max 4 retransmissions
Receiver type MMSE 0.6
Control channel and
CDF

0.5
reference signal overhead Fixed 0.3063
(As agreed in ITU evaluation) 0.4
Simulation time 800 subframes (1 ms/subframe)
0.3

TABLE II 0.2
S YSTEM L EVEL S IMULATION R ESULTS FOR 8T X C O - POLARIZED
A NTENNA 0.1 DFT based codebook
Proposed codebook
0
Codebook (N1 , N2 ) Average SE (bps/Hz) Edge SE (bps/Hz/user) 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
DFT based feedback 4.53 (100%) 0.161 (100%) User Throughput (kbps)
Proposed codebook 4.97 (109.71%) 0.171 (106.21%)
Fig. 2. CDF of user throughput for 8Tx cross-polarized antenna.

IV. S IMULATIONS
This section shows the system level simulation with the increasing the system performance. Furthermore, Table II and
presented codebooks for LTE-A FDD downlink system over III give the performance comparison in the form of spectrum
19 sites/57 pentagon-shaped cells. The main simulation pa- efficiency (SE) between the proposed codebooks in equations
rameters and modeling assumptions for LTE-A evaluation (19)-(22) and DFT based codebooks [6]. It can be seen that
are summarized in Table I. The overall downlink bandwidth with the same overhead, for co-polarized antenna, the proposed
(10 MHz) is divided into 10 subbands, i.e., M = 10. We codebooks greatly outperform DFT based codebooks with
consider perfect channel knowledge at the user side and error- about 10% gain on average cell SE, and 6% gain on cell-
free feedback transmission from each user. For MU-MIMO, edge user SE. And for cross-polarized antenna, the proposed
lower bound of the post-detection signal to interference plus codebooks still show gain compared with DFT based code-
noise ratio (SINR) of each user is used as the channel quality book. Through system evaluations, it is demonstrated that the
indicator (CQI) [7]. After the MU-MIMO scheduling, BS
adjusts the reported CQI according to the number of co- TABLE III
scheduled users by multiplying a scaling factor, and equal S YSTEM L EVEL S IMULATION R ESULTS FOR 8T X C ROSS - POLARIZED
A NTENNA
transmit power is allocated to each data stream.
Figure 1 and figure 2 show the cumulative distribution Codebook (N1 , N2 ) Average SE (bps/Hz) Edge SE (bps/Hz/user)
function (CDF) of user throughput for the proposed code- DFT based feedback 3.64 (100%) 0.114 (100%)
books, which illustrates that the proposed codebooks help Proposed codebook 3.71 (101.92%) 0.117 (102.63%)
proposed codebooks outperform DFT based codebook, while
the overhead is not increased.
V. C ONCLUSIONS
In this paper, a codebook design is proposed by adjust-
ing phase between polarizations in the precoder targeting
wideband CSI. This design effectively matches the channel
characteristics of co-polarized linear antenna array, since the
largest number of DFT beams are constructed for this antenna
configuration. System level simulation in a LTE-A system
is performed to evaluate the performance of the proposed
codebooks. The results show that compared to DFT based
codebooks in [6], without increasing the feedback overhead,
the proposed codebooks show significant gain on both average
cell SE and cell-edge user SE for co-polarized antenna, while
still outperform the latter for cross-polarized antenna.
ACKNOWLEDGMENT
This work is supported by National Science and Tech-
nology Major Projects under Grant 2009ZX03003-010 and
2011ZX03003-001-04.
R EFERENCES
[1] A. Paulraj, R. Nabar and D. Gore, Introduction to Space-Time Wireless
Communications. Cambridge: Cambridge University Press, 2003.
[2] Agilent Technologies, Agilent 3GPP Long Term Evolution: Sys-
tem Overview Developement and Test Challanges, May 2008.
http://cp.literature.agilent.com/litweb/pdf/5989-8139EN.pdf.
[3] E. Seidel, Progress on ”LTE-Advanced” - the new 4G standard, Jul. 2008.
http://www.nomor.de/uploads.
[4] 3GPP TR 36.913 v8.0.1, Requirements for further advancements for
EUTRA (LTE-Advanced), Mar. 2009.
[5] 3GPP, R1-101683, Way forward for Rel-10 Feedback framework, Erics-
son, Huawei, Alcatel-Lucent, Alcatel-Lucent Shanghai Bell, LG Electron-
ics, Marvell, Nokia, Nokia Siemens Networks, NTT DoCoMo, Panasonic,
Philips, Qualcomm Inc., Research In Motion, Samsung, ST-Ericsson,
Texas Instruments, ZTE.
[6] 3GPP, R1-105011, Way forward on 8Tx codebook for Rel.10 DL MIMO,
Alcatel-Lucent, Alcatel-Lucent Shanghai Bell, AT&T, CATT, CEWiT,
CMCC, Ericsson, Kyocera, LG Electronics, LG-Ericsson, Marvell, Mit-
subishi Electric, Motorola, NEC, Nokia, Nokia Siemens Networks, NTT
DoCoMo, Panasonic, Qualcomm Inc., Samsung, Sharp, Sony Corpora-
tion, ST-Ericsson, Texas Instruments.
[7] M. Trivellato, F. Boccardi, F. Tosato, User Selection Schemes for MIMO
Broadcast Channels with Limited Feedback, in Proc. IEEE VTC2007-
Spring, Dublin, Ireland, Apr. 2007, pp. 2089-2093.