See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224147962

Coordinated multipoint transmission/reception techniques for LTE-

Advanced

Article in IEEE Wireless Communications · July 2010

DOI: 10.1109/MWC.2010.5490976 · Source: IEEE Xplore

CITATIONS READS

590 7,107

5 authors, including:

Mamoru Sawahashi Yoshihisa Kishiyama

Tokyo City University NTT DOCOMO
363 PUBLICATIONS 6,134 CITATIONS 238 PUBLICATIONS 8,923 CITATIONS

SEE PROFILE SEE PROFILE

Daisuke Nishikawa
NTT DOCOMO
6 PUBLICATIONS 611 CITATIONS

SEE PROFILE

All content following this page was uploaded by Daisuke Nishikawa on 13 June 2014.

The user has requested enhancement of the downloaded file.

SAWAHASHI LAYOUT 6/9/10 10:39 AM Page 26

C O O R D I N AT E D A N D D I S T R I B U T E D MIMO

COORDINATED MULTIPOINT
TRANSMISSION/RECEPTION TECHNIQUES FOR
LTE-ADVANCED
MAMORU SAWAHASHI, TOKYO CITY UNIVERSITY
YOSHIHISA KISHIYAMA, AKIHITO MORIMOTO, DAISUKE NISHIKAWA, AND MOTOHIRO TANNO,
NTT DOCOMO, INC.

Coherent tr
ABSTRACT respectively [3]. With the circular letter (CL) to
invite proposals for IMT-Advanced radio inter-
This article presents an elaborate coordina- face technologies as the motivation, the 3GPP
tion technique among multiple cell sites called initiated a Study Item (SI, i.e., feasibility study)
eNodeB
coordinated multipoint transmission and recep- for LTE-Advanced in March 2008. In the 3GPP,
tion in the Third Generation Partnership Project LTE-Advanced represents LTE enhancement in
for LTE-Advanced. After addressing major radio Release 10 and beyond. The system require-
(a) access techniques in the LTE Release 8 specifica- ments for LTE-Advanced were agreed upon in
tions, system requirements and applied radio [4]. The technical components to achieve the
Fast sele access techniques that satisfy the requirements system requirements were discussed in the 3GPP,
for LTE-Advanced are described including and summarized in a technical report [5]. Based
CoMP transmission and reception. Then CoMP on Release 8 LTE and agreements on LTE-
eNodeB transmission and reception schemes and the Advanced, the 3GPP submitted its final proposal
related radio interface, which were agreed upon to the ITU-R as a complete submission in Octo-
or are currently being discussed in the 3GPP, are ber 2009. Following the SI, Work Items (WIs,
presented. Finally, system-level simulation evalu- i.e., specification work) for the LTE-Advanced
ations show that the CoMP transmission and radio interface in Release 10 were initiated and
The authors present reception schemes have a significant effect in it is expected that the specifications will be com-
an elaborate terms of improving the cell edge user throughput
based on LTE-Advanced simulation conditions.
pleted according to the IMT-Advanced stan-
dardization schedule. In addition to multiple
coordination access schemes, advanced multiple-input multi-
ple-output (MIMO) channel transmission tech-
technique among INTRODUCTION niques and elaborate coordination among
In the Third Generation Partnership Project multiple cell sites called coordinated multipoint
multiple cell sites (3GPP), the specifications for the Universal (CoMP) transmission/reception were adopted as
Mobile Telecommunications System (UMTS) key techniques for LTE-Advanced at the Tech-
called coordinated Long-Term Evolution (LTE) called the evolved nical Specification Group — Radio Access Net-
multi-point universal terrestrial radio access (UTRA) and uni-
versal terrestrial radio access network (UTRAN)
work (TSG-RAN) Working Group 1 (WG1)
meeting in the 3GPP [5].
transmission and were completed as Release 8 [1, 2]. Release 8 In this article we present the purpose for
LTE supports only efficient packet-based radio CoMP based on the system requirements, and
reception in the access and radio access networks that provide actual CoMP transmission and reception
Internet Protocol (IP)-based functionalities with schemes for LTE-Advanced based on the discus-
3rd generation low latency and low cost. After the completion of sion in the 3GPP. In the rest of the article we
the specifications, commercial equipment is under first address major radio access techniques
partnership project development aiming at the forthcoming launch of including intercell interference coordination
for LTE-Advanced. broadband radio access services.
The principle behind international standard-
(ICIC) in the Release 8 LTE, which form the
basis for the radio interface for LTE-Advanced.
ization of International Mobile Telecommunica- Then we explain the necessity for CoMP trans-
tion (IMT)-Advanced and the new spectra for mission and reception to satisfy the system
IMT were agreed upon at the Radiocommunica- requirements in terms of capacity and cell edge
tion Assembly 2007 (RA ’07) and the World user throughput through tighter intercell orthog-
Radiocommunication Conference 2007 (WRC onality as well as intracell orthogonality among
’07) in the International Telecommunication simultaneously accessing sets of user equipment
Union — Radiocommunication Sector (ITU-R), (UE) in LTE-Advanced. To achieve intercell

26 1536-1284/10/$25.00 © 2010 IEEE IEEE Wireless Communications • June 2010

SAWAHASHI LAYOUT 6/9/10 10:39 AM Page 27

orthogonality, we address two types of intercell SYSTEM REQUIREMENTS AND The target values in
radio resource management configurations: cen-
tralized and autonomous. In particular, CoMP APPLIED TECHNIQUES FOR LTE-ADVANCED terms of the peak
transmission and reception provide a significant LTE-Advanced has two-faceted general require-
gain in terms of capacity and cell edge user ments. The first facet is that LTE-Advanced shall data rate and peak
throughput in centralized tight intercell radio be an evolution of Release 8 LTE. This means
resource management, which takes advantage of that LTE-Advanced should achieve much higher spectrum efficiency
remote radio equipment (RRE), that is, a levels of system performance than those for
remote base station (BS). Finally, we present Release 8 LTE to satisfy the increasing future will be mainly
system-level simulation results on gains of CoMP
transmission in the downlink and reception in
traffic demand. Moreover, LTE-Advanced must
be fully backward compatible with Release 8
achieved by extend-
the uplink assuming the LTE-Advanced simula- LTE to enable continuous enhancement and ing the transmission
tion conditions of the 3GPP. deployment. The second facet to the require-
ments is that LTE-Advanced shall meet or exceed bandwidth to around
the IMT-Advanced requirements within the ITU-
SYSTEM REQUIREMENTS RELATED TO R time plan. Based on the general requirements, 100 MHz and
COMP TRANSMISSION AND RECEPTION the capability-related requirements and system
increasing the
requirements focusing on physical layer perfor-
RELEASE 8 LTE mance are summarized in Table 1. Based on the number of
description in the CL referring to ITU-R Recom-
The system requirements for Release 8 LTE are mendation M.1645, the target peak data rate in transmitter and
specified in [6]. The performance evaluation the downlink is 1 Gb/s for LTE-Advanced. The
results for Release 8 LTE are given in [7]. They target peak data rate in the uplink is set to 500 receiver antennas
show that the system requirements are satisfied Mb/s considering the future traffic demand in
and validate the radio interface in the Release 8 cellular networks. The achievement values in MIMO channel
LTE specifications. In the Release 8 LTE, regarding the peak spectrum efficiency for
orthogonal frequency-division multiple access Release 8 LTE are 15 and 3.75 b/s/Hz in the transmission.
(OFDMA) and single-carrier frequency-division downlink and uplink, respectively. They are
multiple access (SC-FDMA) based on discrete achieved by using four-stream or single-stream
Fourier transform (DFT)-spread orthogonal fre- transmission, respectively, both with 64-quadra-
quency-division multiplexing (OFDM) are adopt- ture amplitude moduation (QAM). Taking into
ed in the downlink and uplink, respectively. account the increasing traffic demand and spec-
Hence, intracell orthogonality among simultane- trum availability when LTE-Advanced is
ously accessing UE sets is achieved in the time deployed, further improvement in the peak spec-
and frequency (and code) domains for both trum efficiency is necessary. Accordingly, the tar-
links. Intercell macro diversity is not applied in get values in terms of the peak spectrum
either link. In the downlink a distinct macro efficiency are set to 30 and 15 b/s/Hz in the
diversity gain is not obtained for the physical downlink and uplink, respectively. It should be
downlink shared channel (PDSCH) due to the noted that these targets are not mandatory for all
slight gain caused by mutual interference, which UE categories, and are to be achieved by a com-
is similar to the high-speed physical downlink bination of BSs and high-class UE with a larger
shared channel (HS-PDSCH) in High Speed number of antennas. The target values in terms
Downlink Packet Access (HSDPA) in intercell of the peak data rate and peak spectrum efficien-
asynchronous cell site operation. It is reported in cy will mainly be achieved by extending the trans-
[8] that a distinct macro diversity effect is gained mission bandwidth to around 100 MHz and
in the uplink even with channel-dependent increasing the number of transmitter and receiv-
scheduling and hybrid automatic repeat request er antennas in MIMO channel transmission
(HARQ) with soft combining. However, intercell (maximum of eight and four antennas in the
macro diversity is not adopted in the uplink downlink and uplink, respectively, were approved
either. This is because the single-node (i.e., as optional usage for LTE-Advanced).
eNode B) architecture is required to achieve the The capacity corresponds to the average cell
required short control and transmission delays spectrum efficiency. The cell edge user through-
and efficient radio resource management, and put is defined as the 5 percent value in the cumu-
reduce implementation and deployment costs. lative distribution function (CDF) of the user
Moreover, a radio interface pertaining to throughput assuming 10 UE sets with a full
FDMA-based ICIC using backhaul signaling buffer per cell. Both are very important require-
with semi-static control was introduced to reduce ments from the system performance viewpoint.
the influence of other-cell interference in In addition to the requirements for IMT-
Release 8 LTE. However, intercell randomiza- Advanced, the 3GPP defined its own targets for
tion is the baseline in Release 8 LTE, and the LTE-Advanced assuming the same evaluation
introduction of ICIC employing partial frequen- scenario that was used for Release 8 LTE (case
cy reuse at the cell boundary is an implementa- 1 in [5]) as shown in Table 1. Since LTE-
tion matter for the operator. This is because the Advanced will support BSs and UE with various
system requirements, including capacity and cell antenna configurations, the target figures in
edge user throughput, are satisfied mainly due to terms of capacity and cell edge user throughput
intracell orthogonal multi-access and frequency were defined for each antenna configuration. In
domain channel-dependent scheduling associat- LTE-Advanced 1.4 to- 1.6-fold improvements in
ed with mandatory two-branch antenna diversity capacity and cell edge user throughput are
reception at the UE without intercell macro expected from Release 8 LTE except for increas-
diversity and ICIC. ing the number of antennas.

IEEE Wireless Communications • June 2010 27

SAWAHASHI LAYOUT 6/9/10 10:39 AM Page 28

Since LTE-Advanced Downlink/ Antenna Rel. 8 LTE

LTE-Advanced IMT-Advanced
Uplink configuration achievement
will support BSs
and UEs with Peak data rate
Downlink — 300 Mb/s 1 Gb/s
1 Gb/s
various antenna Uplink — 75 Mb/s 500 Mb/s

configurations, the Peak spectrum

Downlink — 15 30 15

target figures in efficiency [b/s/Hz]

Uplink — 3.75 15 6.75
terms of the capacity 2×2 1.69 2.4 —
and cell-edge user
Downlink 4×2 1.87 2.6 2.2*
throughput were Capacity
4×4 2.67 3.7 —
defined for [b/s/Hz/cell]

each antenna Uplink

1×2 0.74 1.2 —

configuration. 2×4 — 2.0 1.4*

2×2 0.05 0.07 —

Downlink 4×2 0.06 0.09 0.06*

Cell edge user
throughput 4×4 0.08 0.12 —
[b/s/Hz/cell/user]
1×2 0.024 0.04 —
Uplink
2×4 — 0.07 0.03*

*Required values in base coverage urban test environment

Table 1. System performance requirements for LTE-Advanced.

Here, we consider techniques that will satisfy tical to that for Release 8 LTE. The peak data
the requirements for capacity and cell edge user rate increases by increasing the number of CCs
throughput. In LTE-Advanced the concept of and that of the transmitter or receiver antennas
carrier aggregation is adopted to achieve a wider in MIMO multiplexing. The peak spectrum effi-
transmission bandwidth up to approximately 100 ciency can be improved mainly by increasing the
MHz [5]. In carrier aggregation the entire trans- number of transmitter or receiver antennas as
mission bandwidth comprises multiple frequency well. However, to satisfy the requirements on
blocks called component carriers (CCs). The capacity and cell edge user throughput, multius-
bandwidth of a CC corresponds to the transmis- er MIMO transmission and intercell interference
sion bandwidth of the system bandwidth speci- control through intercell orthogonal radio
fied in Release 8 LTE to enable the radio link resource assignment among simultaneous access-
connection for Release 8 LTE UE within the ing users are necessary. In particular, intercell
same spectrum. Moreover, to achieve flexible orthogonal radio resource assignment is recog-
spectrum utilization, it was decided that carrier nized as a key enhancement from the Release 8
aggregation shall support both contiguous and LTE based on intracell orthogonal radio access
non-contiguous spectra. Note that the non-con- as shown in Table 2, and has been actively inves-
tiguous spectrum usage in the same link is called tigated for CoMP transmission and reception for
spectrum aggregation. Multiple access schemes LTE-Advanced in the 3GPP associated with
for LTE-Advanced are decided based on the multiple antenna implementations at a BS and
concept that higher priority is given to backward UE. The CoMP mainly contributes to increasing
compatibility and easy implementation rather the cell edge user throughput (which means it
than performance gain (e.g., by channel-depen- can also extend the effective coverage).
dent scheduling). Actually, OFDMA with a CC-
based structure is adopted in the downlink, and
N times, i.e., CC based, SC-FDMA using DFT- FAST INTERCELL RADIO RESOURCE
Spread OFDM is adopted in the uplink (N indi- MANAGEMENT SCHEMES FOR
cates the number of CCs in the transmission
bandwidth) [5]. Hence, the frequency span for COMP TRANSMISSION AND RECEPTION
frequency domain channel-dependent scheduling To achieve tight intercell orthogonal radio
and frequency diversity with distributed trans- resource assignment, fast intercell radio resource
mission is confined within one CC for one trans- management (or, equivalently, interference man-
port block. Thus, the achievable frequency agement) is necessary. Figure 1 illustrates the
diversity gain for LTE-Advanced is almost iden- radio access network structure for fast intercell

28 IEEE Wireless Communications • June 2010

SAWAHASHI LAYOUT 6/9/10 10:39 AM Page 29

W-CDMA Rel. 8 LTE LTE-Advanced

Downlink (Partially) orthogonal Orthogonal Orthogonal

Intracell
Uplink Non-orthogonal Orthogonal Orthogonal

Downlink Non-orthogonal Non-orthogonal (Quasi)-orthogonal

Intercell
Uplink Non-orthogonal Non-orthogonal (Quasi)-orthogonal

Table 2. Intra- and inter-cell orthogonality in radio access systems.

radio resource management schemes that

achieve CoMP transmission and reception. As Inter-eNode B (X2) I/F
shown in the figure, two approaches are consid- Centralized control
Autonomous control
ered for fast intercell radio resource manage- Baseband signaling
over optical fiber
ment: centralized and autonomous. eNodeB eNodeB

Centralized Intercell Radio Resource Management — As RRE RRE

RRE
shown in Fig. 1, we deploy a group of RREs, RRE RRE
which are connected to the central BS (eNode
B) by an optical fiber. Hence, complete coordi-
nated transmission or reception is achieved Figure 1. Radio network structure for fast inter-cell radio resource management.
among RREs through unified radio resource
management at the central BS. Since the inter-
face between the central BS and RRE can be both the downlink and uplink is not necessarily
designed within one transceiver, fast transfer of optimum from the viewpoints of minimum trans-
the baseband digital signal is possible. RREs are mission power and reduction in interference. In
effectively used in the existing second- and third- this case, nevertheless, the optimum independent
generation networks by installing small BSs in link connection to different RREs is possible
limited space and reducing the cable loss. Hence, between the downlink and uplink by centralized
centralized intercell radio resource management intercell radio resource management. That is, the
employing RREs is deployed more aggressively uplink can be connected to the RRE of the pico-
to achieve CoMP transmission and reception cell, while the downlink is connected to the RRE
among different cell sites for LTE-Advanced. of the macro/microcell. By having centralized
radio resource management utilize the RREs,
Autonomous Intercell Radio Resource Management — the scheduling and ARQ operations are achieved
This method was already introduced into in a unified manner at the central BS. Moreover,
Release 8 LTE using the X2 interface (i.e., inter- ICIC with autonomous intercell radio resource
eNode B interface). The X2 interface is mainly management is being discussed aggressively in
used for inter-eNode B handover, that is, to the 3GPP as an essential technique to support
transfer the UE context in the control plane and heterogeneous cell configuration. The gains of
forward user data in the user plane using IP ICIC using the X2 interface are presented for
transport. The X2 interface is also used for mul- mitigating the interference between eNode Bs in
ticell radio resource management functions macrocells and home eNode Bs in femtocells
including ICIC employing semi-static partial fre- which may provide restricted access only to UE
quency reuse. It was reported in [9] that the typ- belonging to the allowed group, called a closed
ical average delay in the X2 interface via the subscriber group (CSG).
backhaul is expected to be in the neighborhood
of 10 ms. Thus, the radio interface must be
enhanced with respect to the delay in both the COMP TRANSMISSION AND
control and user planes via the backhaul and/or RECEPTION SCHEMES FOR LTE-ADVANCED
air to achieve faster radio resource management
among different cell sites for autonomous inter- DOWNLINK
cell radio resource management.
According to the deployment environment, Transmission Scheme — In downlink CoMP trans-
fast centralized radio resource management, mission, two transmission schemes are mainly
autonomous intercell radio resource manage- considered: joint processing (JP) and coordinat-
ment, or a combination of these is applied. In ed scheduling/beamforming (CS/CB) [10, 11]. In
particular, centralized intercell radio resource CoMP transmission the related control channels,
management using RRE is very effective, and including the physical downlink control channel
assumed to achieve CoMP transmission and (PDCCH), are transmitted only from the serving
reception. One of the promising applications of (anchor) cell regardless of the transmission
coordinated multicell transmission is efficient scheme [5].
support of heterogeneous networks. In a hetero-
geneous cell configuration such as when Joint Processing — JP is further categorized into
macro/microcells coexist with picocells within the joint transmission (JT) and dynamic cell selec-
cell area, link connection to the same cell site in tion (DCS). Figure 2a shows the operating prin-

IEEE Wireless Communications • June 2010 29

SAWAHASHI LAYOUT 6/9/10 10:39 AM Page 30

Coordinated Scheduling/Beamforming — Figure 2c

Coherent transmission shows the operating principle of CB (note that
CS corresponds to the case without beamform-
ing). Unlike the aforementioned DCS, an RB of
the PDSCH is transmitted only from the serving
eNodeB RRE cell together with the PDCCH. Hence, an RB is
Not scheduled
assigned to the UE with CS/CB by scheduling of
the serving cell. However, scheduling/beamform-
(a) ing is coordinated among multiple coordinated
cells. In this case transmit beamforming weights
Fast selection for each UE set are generated to reduce the
unnecessary interference to other UE scheduled
within the coordinated cells. Therefore, in par-
eNodeB RRE ticular, the cell edge user throughput can be
improved due to the increase in received SINR.
Not scheduled As we mention later, since the downlink
(b)
CoMP transmission mainly contributes to
improving the cell edge user throughput, more
Interference coordination advanced CoMP transmission schemes combined
beamforming with MIMO multiplexing (particularly multiuser
MIMO) techniques were proposed to improve
the capacity and its potential gain has been
eNodeB reported in many papers [10, 11].
RRE

Radio Interface to Support Downlink CoMP Transmission

— In Release 8 LTE a cell-specific RS (CS-RS)
(c)
for up to four antennas is specified and used for
channel state information (CSI) measurement
Figure 2. CoMP transmission in downlink: a) joint transmission (JT); b) for channel-dependent scheduling, handover,
dynamic cell selection (DCS); c) coordinated beamforming. and demodulation for coded data symbols. In
LTE-Advanced two types of RSs are specified in
addition to the Release 8 CS-RS for the purpose
ciple of JT in the downlink. In JT the same of CoMP transmission and to support more
resource block (RB) of the PDSCH is transmit- antennas than four and up to eight in MIMO
ted from multiple cells associated with a UE- multiplexing: CSI-RS and US-RS. The CSI-RS is
specific demodulation reference signal (US-RS) transmitted sparsely (e.g., longer than a 5 ms
among coordinated cells (i.e., from non-serving duration) to achieve low overhead. The US-RS
cell(s) as well as the serving cell). For instance, is transmitted with the same precoding scheme
JT is achieved by codebook-based precoding to as that for the PDSCH. Hence, channel estima-
reduce feedback signal overhead. In principle, tion and demodulation in a unified manner is
the best precoding matrixes for intercell site possible irrespective of the CoMP transmission
coordination are selected in addition to the scheme. Moreover, explicit CSI feedback (i.e.,
individual selection of the best precoding matrix direct feedback of the channel gain) is investi-
at each cell site so that the received signal-to- gated to conduct precise precoding, in addition
interference-plus-noise power ratio (SINR) is to implicit CSI feedback including the precoding
maximized at a UE set among the predeter- matrix index (PMI) based on Release 8 LTE. In
mined precoding matrix candidates. Other this case the key factor is the trade-off between
implementation methods are also considered to the achievable gain and the feedback signaling
achieve the principle operation. A UE set feeds overhead.
back a channel quality indicator (CQI) based on
the combined received SINR to the serving cell, UPLINK
and then an RB is dynamically assigned to the In CoMP reception in the uplink, the physical
UE by fast scheduling at the central BS. Since uplink shared channel (PUSCH) is received at
the transmission power resources of multiple multiple cells. In this case, maximal ratio com-
cell sites can be used through coherent trans- bining (MRC) is used at multiple RREs. Figures
missions, the cell edge user throughput is 3a and 3b show the CoMP reception with inter-
improved significantly. Figure 2b shows the ference rejection combining (IRC) and that with
operating principle of DCS. In DCS an RB of coordinated scheduling, respectively. As shown
the PDSCH associated with a US-RS is trans- in Fig. 3a, multiple UE sets transmit the PUSCH
mitted from one cell among the coordinated simultaneously using the same RB; however,
cells, and the cell transmitting the PDSCH with received weights are generated so that the
the minimum path loss is dynamically selected received SINR or signal power after combining
through fast scheduling at the central BS. Then, at the central eNode B is maximized in CoMP
for instance, the other cells among the coordi- reception with IRC. The minimum mean squared
nated cells are muted (i.e., they do not transmit error (MMSE) or zero forcing (ZF) algorithm is
the RB), so the cell edge UE does not receive typically used to combine the received PUSCHs
other-cell interference. Therefore, the maximum at multiple cell sites. As shown in Fig. 3b, only
received signal power is obtained, and the inter- one UE set transmits the PUSCH using an RB
ference from neighboring cells is significantly based on the coordinated scheduling among cells
mitigated. in CoMP reception with coordinated scheduling.

30 IEEE Wireless Communications • June 2010

SAWAHASHI LAYOUT 6/9/10 10:39 AM Page 31

System bandwidth 10 MHz

Cellular layout Hexagonal grid, 19 cell sites, 3 cells per cell site with 70 degree sector beam width

Inter-site distance 500 m

Number of UEs per cell 10 (randomly assigned with a uniform distribution)

Minimum distance between UE and cell 35 m

Distance dependent path loss 128.1 + 37.6log10(r) dB, r in kilometers

Shadowing standard deviation 8 dB

Shadowing correlation 0.5 (inter-site)/1.0 (intra-site)

Penetration loss 20 dB

Channel model 6 ray typical urban

Maximum Doppler frequency fD = 5.55 Hz (3 km/h @ 2 GHz)

eNode B/UE Tx power 46 dBm/23 dBm

eNode B/UE antenna gain 14 dBi/0 dBi

eNode B/UE antenna height 32 m/1.5 m

eNode B vertical antenna pattern 10 degree vertical beam width with tilt angle of 15 degrees

Front-to-back ratio of eNode B antenna 25 dB

Antenna configurations Downlink: 2-by-2, 4-by-2, Uplink: 1-by-2, 1-by-4

Scheduling algorithm Proportional fairness

Control delay (scheduling, AMC, downlink

6 msec
PMI selection, and cell selection for CS)

Modulation and coding schemes QPSK (R = 1/8–5/6), 16-QAM (R = 1/2–5/6), 64-QAM (R = 3/5–4/5)

Hybrid ARQ Chase combining (8 msec round trip delay for retransmission)

Traffic model ON/OFF traffic (activity factor parameterized)

Table 3. Simulation conditions for LTE-Advanced in 3GPP.

In both schemes, the cell-edge user throughput neous effective SINR over one transmission time
is improved due to the increase in the received interval (TTI). In the system-level simulation the
signal power. Note that CoMP reception in the cell throughput and cell edge user throughput
uplink is an implementation matter and does not are calculated by adding random errors accord-
require a significant change in the physical layer ing to the mapping between the measured effec-
radio interface. tive SINR and the BLER performance derived
from the link-level simulation. We employ the
SIMULATION EVALUATIONS exponential effective SIR mapping (EESM)
method in the combined link- and system-level
SIMULATION CONFIGURATION simulations. Table 3 lists the major parameters
in the link- and system-level simulations, which
We evaluate the cell throughput and cell edge basically correspond to those agreed upon in the
user throughput performance applying CoMP 3GPP for LTE-Advanced [5].
transmission and reception schemes by combin-
ing link-level and system-level simulations. More Link-Level Simulation Parameters — The RB band-
specifically, in the link-level simulation we mea- width is 180 kHz with 12 subcarriers. The sub-
sure the average block error rate (BLER) of band bandwidth is 1.08 MHz (= 6 RBs) in the
each modulation and coding scheme (MCS) downlink. One 1-ms-long TTI contains 14
associated with turbo coding and iterative soft- OFDM symbols, each of which comprises a 66.7
decision turbo decoding against the instanta- μs effective symbol and a 4.7 μs cyclic prefix

IEEE Wireless Communications • June 2010 31

SAWAHASHI LAYOUT 6/9/10 10:39 AM Page 32

SC-FDMA symbols for demodulation. More-

MMSE, ZF over, a Zadoff-Chu sequence is used for the
DM-RS. The channel gain is computed from a
coherent averaging of two DM-RS symbols at
each TTI. A frequency domain equalizer based
eNodeB RRE
on the MMSE is used. One sounding reference
signal (SRS) symbol multiplexed into the last
SC-FDMA symbol per two TTIs is considered as
(a)
the overhead of CSI measurement for schedul-
ing.
MRC

System-Level Simulation Parameters — The entire

system bandwidth is 10 MHz (the occupied band-
eNodeB width is 9 MHz, which corresponds to 50 RBs).
RRE
We assume a carrier frequency of 2.0 GHz. A 3-
No transmission sectored 19-hexagonal cell site model is assumed
with a sector antenna beam pattern with a 70˚
(b) beam width (opening angle), which is defined as
a beam width with the gain 3 dB lower than the
Figure 3. CoMP reception in uplink: a) multipoint reception with interference peak value [13]. The propagation model follows
rejection combining; and b) multipoint reception with coordinated scheduling. distance-dependent path loss with a decay factor
of 3.76, lognormal shadowing with the standard
deviation of 8 dB, and instantaneous multipath
(CP). The modulation schemes are quaternary fading. For each trial, it is assumed that the dis-
phase shift keying (QPSK), 16-QAM, and 64- tance-dependent path loss is constant during the
QAM in both the downlink and uplink. We use throughput measurement period, while the time-
turbo coding with the constraint length of four varying shadowing and instantaneous fading vari-
bits using the variable coding rates given in ations are added. We employ proportional
Table 3. In the downlink two- or four-antenna fairness-based time and frequency domain chan-
transmission and two-antenna diversity reception nel-dependent scheduling, and adaptive modula-
are assumed. In the case of single-cell transmis- tion and coding (AMC) for instantaneous
sion we use the CS-RS for two or four transmis- channel variation. We apply Chase combining as
sion antennas used for CSI measurement and HARQ with packet combining. We employ the
demodulation specified in the Release 8 LTE ON/OFF traffic model. In the ON/OFF traffic
[12]. In the case of JT and fast CS, we employ model, TTIs scheduled to a UE set are randomly
the CSI-RS and US-RS specified in LTE- assigned, and the ratio of the assigned TTIs to
Advanced [5]. In this article we assume that US- the entire time duration is parameterized as the
RSs are mapped every six subcarriers onto the activity factor.
fourth, seventh, tenth, and thirteenth OFDM In the downlink we assume that the CQI after
symbols. In the uplink one-antenna transmission coherent combining in JT is ideally measured,
and two or four-antenna diversity reception are and the CQI and PMI are fed back to the eNode
assumed. The demodulation RS (DM-RS) sym- B without decoding error. Furthermore, we
bols are multiplexed into the fourth and eleventh select the best precoding matrix for each cell site

2 × 2 downlink 4 × 2 downlink
12 Joint transmission 12 Joint transmission
Coordinated scheduling Coordinated scheduling
Single-cell transmission Single-cell transmission
10 10
Cell edge user throughput (Mb/s)

Cell edge user throughput (Mb/s)

8 8

6 6

4 4

2 2

0 0
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Cell throughput (Mb/s) Cell throughput (Mb/s)

(a) (b)

Figure 4. Cell edge user throughput as function of cell throughput in downlink: a) 2 × 2; b) 4 × 2.

32 IEEE Wireless Communications • June 2010

SAWAHASHI LAYOUT 6/9/10 10:39 AM Page 33

1 × 2 uplink 1 × 4 uplink
MMSE MMSE
ZF ZF
1.6 Coordinated scheduling 1.6 Coordinated scheduling
Single-cell reception Single-cell reception

Cell edge user throughput (Mb/s)

Cell edge user throughput (Mb/s)

1.2 1.2

0.8 0.8

0.4 0.4

0 0
0 4 8 12 16 0 4 8 12 16
Cell throughput (Mb/s) Cell throughput (Mb/s)

(a) (b)

Figure 5. Cell edge user throughput as function of cell throughput in uplink: a) 1 × 2; b) 1 × 4.

individually and for inter-cell-site coordination al loaded cases. For instance, when the cell
among the four precoding matrix candidates throughput of 12 Mb/s is achieved in the 2 × 2
defined in Release 8 LTE; that is, equivalently antenna configuration case, the cell edge user
select the best precoding matrix among the 4 × 4 throughput is increased approximately 58 and 43
× 4 candidates, and the best cell subband by sub- percent compared to the single-cell transmission
band. The cell edge UE to which CoMP trans- by applying JT and fast CS, respectively. More-
mission is applied is defined to be when the over, JT achieves higher cell edge user through-
difference in the UE path loss between the serv- put than fast CS, since it fully utilizes the higher
ing cell and the second strongest cell is within 3 signal power from multiple cell sites through
dB. We assume single-user CoMP transmission coherent combining at the UE receiver. The rel-
where an RB is scheduled to only one UE in the ative improvement in the cell edge user through-
downlink. As the overhead, we assume the put with JT or fast CS compared to that for
PDCCH of two OFDM symbols per subframe. single-cell transmission for a 4 × 2 antenna con-
In the uplink we apply fractional transmission figuration is almost identical to that for a 2 × 2
power control with α = 0.6 and P0 = –60 dBm, configuration. However, the cell edge user
where α and P0 indicate the attenuation factor throughput for the 4 × 2 configuration is further
of the target received signal level and the target improved by approximately 34 and 30 percent
received signal level when the path loss from the when using JT and fast CS, respectively, com-
BS is zero, respectively. For CoMP reception in pared to that for the 2 × 2 configuration due to
the uplink, we assume combined reception the increasing precoding gain.
between two cells with the highest and second Figures 5a and 5b show the cell edge user
highest average received power. We assume the throughput performance in the uplink as a func-
PUCCH of 10 RBs as the overhead. tion of the cell throughput with 1 × 2 and 1 × 4
antenna configurations, respectively. We plot the
SIMULATION RESULTS performance of the CoMP reception schemes
Figures 4a and 4b show the cell edge user employing MMSE combining, ZF combining,
throughput performance in the downlink as a and MRC with coordinated scheduling, and that
function of the cell throughput assuming 2 × 2 of single-cell reception as a reference. The fig-
and 4 × 2 antenna configurations, respectively. ures show that similar to the downlink CoMP
We plot the performance of the CoMP transmis- transmission, CoMP reception increases the cell
sion schemes employing JT and fast CS, and sin- edge user throughput when the same cell
gle-cell transmission as a reference. In the throughput is achieved in the fractional loaded
figures the activity factor of the ON/OFF traffic cases. For instance, when the cell throughput of
model is parameterized to obtain each plot. 8 Mb/s is achieved with a two-receiver-antenna
More specifically, the plot achieving the highest configuration, the cell edge user throughput is
cell throughput for each transmission scheme increased approximately 54, 38, and 16 percent
corresponds to the full load case with an activity compared to single-cell reception by applying
factor of one, while the plots for the lower cell the CoMP reception schemes using MMSE com-
throughput values are obtained in the fractional bining, ZF combining, and MRC with coordinat-
loaded cases with lower activity factors. The fig- ed scheduling, respectively. Moreover, the CoMP
ures show that the CoMP transmission schemes reception scheme with MMSE or ZF combining
increase the cell edge user throughput when the achieves higher cell throughput in a full load
same cell throughput is achieved in the fraction- case since these schemes do not require coordi-

IEEE Wireless Communications • June 2010 33

SAWAHASHI LAYOUT 6/9/10 10:39 AM Page 34

The system-level nated scheduling, which has a negative impact

on the cell throughput. For instance, CoMP
[7] 3GPP TR 25.912, V8.0.0, “Feasibility Study for Evolved
Universal Terrestrial Radio Access (UTRA) and Universal
Terrestrial Radio Access Network (UTRAN),” Dec. 2008.
simulation results reception with MMSE (ZF) combining increases [8] H. Kawai et al., “Investigations on Inter-Node B Macro
the cell throughput in a full load case by approx- Diversity for Single-Carrier Based Radio Access in
showed that the imately 19 (11) and 22 (19) percent with a two- Evolved UTRA Uplink,” Proc. IEEE Sarnoff Symp., May
2007.
or four-receiver-antenna configuration, respec-
CoMP transmission tively. This is due to the decreasing noise
[9] 3GPP R3-070702, “Reply LS to R3-070527/R1-071242
on Backhaul (X2 interface) Delay,” Mar. 2007.
enhancement in MMSE or ZF combining
in the downlink and according to the increase in the number of
[10] M. K. Karakayali, G. J. Foschini, and R. A. Valenzuela,
“Network Coordination for Spectrally Efficient Commu-
nications in Cellular Systems,” IEEE Wireless Commun.,
reception in the receiver antennas at the BS. vol. 13, no. 4, Aug. 2006.
[11] J. G. Andrews, W. Choi, and R. W. Heath Jr., “Over-
uplink are very CONCLUSION coming Interference in Spatial Multiplexing MIMO Cel-
lular Networks,” IEEE Wireless Commun., vol. 14, no. 6,
effective in improv- This article provides an overview of the system Dec. 2007, pp. 95–104.
[12] 3GPP TS 36.211, V9.1.0, “Evolved Universal Terrestrial
requirements and applied radio access tech- Radio Access (E-UTRA); Physical Channels and Modula-
ing the cell-edge niques that satisfy the requirements, including tion,” Mar. 2010.
CoMP transmission and reception, for LTE- [13] B. Christer, V. Johansson, and S. Stefansson, “Optimiz-
user throughput. Advanced. Moreover, we present fast intercell ing Antenna Parameters for Sectorized W-CDMA Net-
works,” Proc. of IEEE VTC-Fall, vol. 4, Sept. 2000, pp.
Considering actual radio resource management schemes that
achieve elaborate CoMP, and the CoMP trans-
1524–31.
[14] A. Furuskär, “Performance Evaluations of LTE-Advanced
application, further mission and reception schemes including the — The 3GPP ITU Proposal,” Proc. WPMC ‘09, Sept.
2009.
related radio interface agreed upon by the 3GPP
investigation is for LTE-Advanced. The system-level simulation
results show that CoMP transmission in the BIOGRAPHIES
necessary. downlink and reception in the uplink are very MAMORU SAWAHASHI [M] (sawahasi@tcu.ac.jp) received his
B.S. and M.S. degrees from Tokyo University in 1983 and
effective in improving the cell edge user through- 1985, respectively, and his Dr.Eng. degree from the Nara
put. Considering actual application, further Institute of Technology in 1998. While at NTT and NTT
investigation is necessary on the influence of DOCOMO, he was engaged in the research and develop-
timing error and the propagation time delay. ment of radio access technologies for W-CDMA, and
broadband packet radio access technologies for 3G long-
Along with a wider transmission bandwidth com- term evolution and systems beyond IMT-2000. In April
prising multiple CCs, enhanced multiple access 2006 he assumed the position of professor with the
schemes, and advanced MIMO channel tech- Department of Information Network Engineering, Tokyo
niques including multiuser MIMO transmission, City University.
CoMP transmission and reception will provide Y OSHIHISA K ISHIYAMA received his B.E., M.E., and Dr.Eng.
broadband packet radio access with much higher degrees from Hokkaido University, Sapporo, Japan, in
performance than those for Release 8 LTE 1998, 2000, and 2010,respectively. In 2000 he joined NTT
regarding spectrum efficiency, capacity, and cell DOCOMO, Inc. His research interests include radio access
technologies for 3G long-term evolution and systems
edge user throughput. It should be noted that beyond IMT-2000. He was a recipient of the 2003 Active
the LTE-Advanced radio interface using multi- Research Award in Radio Communication Systems from the
ple access schemes with the presented key tech- IEICE. He was a recipient of the IEICE Young Engineer
niques satisfies the system requirements for Award in 2004.
IMT-Advanced [14]. AKIHITO MORIMOTO received his B.E. and M.E. degrees in infor-
mation electronics from Nagoya University, Japan, in 1997
REFERENCES and 1999, respectively, and received his Dr. Eng. degree
[1] 3GPP TS 36.300, V8.9.0, “Evolved Universal Terrestrial from Tohoku University, Sendai, Japan, in 2008. In 1999 he
Radio Access (EUTRA) and Evolved Universal Terrestrial joined NTT Mobile Communications Network, Inc. (now NTT
Radio Access Network (EUTRAN); Overall Description,” DOCOMO, Inc.). His research interests include mobile com-
June 2009. munication systems. He is a member of the IEICE.
[2] D. Astely et al., “LTE: The Evolution of Mobile Broad-
band,” IEEE Commun. Mag., vol. 47, no. 4, Apr. 2009, DAISUKE NISHIKAWA received his B.E. and M.E. degrees from
pp. 44–51. Kyoto University, Kyoto, Japan in 2004 and 2006, respec-
[3] Final Acts WRC ‘07, Geneva, Nov. 2007. tively. In 2006, he joined NTT DOCOMO, Inc. His research
[4] 3GPP TR 36.913, V9.0.0, “Requirements for Further interests include mobile communication systems. He is a
Advancements for Evolved Universal Terrestrial Radio member of the IEICE.
Access (E-UTRA) (LTE-Advanced),” Dec. 2009.
[5] 3GPP TR 36.814, V9.0.0, “Further Advancements for E- M OTOHIRO T ANNO received his B.E. and M.E. degrees from
UTRA Physical Layer Aspects,” Mar. 2010. Kyoto University, Japan in 1993 and 1995, respectively. In
[6] 3GPP TR 25.913, V8.0.0, “Requirements for Evolved 1995 he joined NTT Mobile Communications Network, Inc.
UTRA (E-UTRA) and Evolved UTRAN (E-UTRAN),” Dec. (now NTT DOCOMO, Inc.) His research interests include
2008. mobile communication systems. He is a member of the IEICE.

34 IEEE Wireless Communications • June 2010

View publication stats