-
Hindawi Publishing CorporationEURASIP Journal on Wireless
Communications and NetworkingVolume 2010, Article ID 161642, 14
pagesdoi:10.1155/2010/161642
Research Article
ANext GenerationWireless Simulator Based onMIMO-OFDM:LTE Case
Study
Gerardo Gómez, DavidMorales-Jiménez, Juan J.
Sánchez-Sánchez,and J. Tomás Entrambasaguas
Department of Communications Engineering, University of Malaga,
29071 Malaga, Spain
Correspondence should be addressed to Gerardo Gómez,
[email protected]
Received 1 June 2009; Revised 29 September 2009; Accepted 3
February 2010
Academic Editor: Faouzi Bader
Copyright © 2010 Gerardo Gómez et al. This is an open access
article distributed under the Creative Commons AttributionLicense,
which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properlycited.
The complexity of next generation wireless systems is growing
exponentially. The combination of Multiple-Input
Multiple-Output(MIMO) technology with Orthogonal Frequency Division
Multiplexing (OFDM) is considered as the best solution to
providehigh data rates under frequency-selective fading channels.
The design and evaluation of MIMO-OFDM systems require a
detailedanalysis of a number of functionalities such as MIMO
transmission modes, channel estimation, MIMO detection, channel
coding,or cross-layer scheduling. In this paper we present a
MIMO-OFDM-based simulator that includes the main physical and link
layerfunctionalities. The simulator has been used to evaluate the
performance of the 3GPP Long-Term Evolution (LTE) technology
fordifferent MIMO-OFDM techniques under realistic assumptions such
as user mobility or bandwidth-limited feedback channel.
1. Introduction
Orthogonal Frequency Division Multiplexing (OFDM) isone of the
most popular physical layer technologies forcurrent broadband
wireless communications due to its highspectral efficiency and
robustness to frequency selectivefading. The use of Multiple-Input
Multiple-Output (MIMO)technology in combination with OFDM increases
the diver-sity gain and/or the system capacity by exploiting
spatialdomain. Hence, MIMO-OFDM is an attractive solution forfuture
broadband wireless systems like 3GPP Long-TermEvolution (LTE) [1,
2] and it will surely be a serious candidatefor future 4G
technologies.
The evaluation of the physical layer performance
under“realistic” varying channel conditions requires
extremelytime-consuming simulations. The use of specialized
math-ematical tools such as MATLAB/Simulink or Mathematicais widely
extended to perform such simulations. These toolsprovide a wide set
of built-in libraries that allow a rapiddevelopment of prototypes.
There are other simulation toolslike Visual System Simulator (VSS)
or Ptolemy II that includestandard building blocks to model and
simulate complexcommunication systems.
However, the performance evaluation of LTE technol-ogy requires
specific MIMO-OFDM-based simulators tominimize the simulation time.
An open-source MATLAB-based LTE physical layer simulator is
presented in [3];nevertheless, it is proved that general purpose
simulationplatforms like MATLAB lead to very long execution times
[4].Other works are only focused on the LTE simulation results[5,
6] under different MIMO-configurations, but no detailson the
employed simulator are provided. Besides, realisticassumptions such
as user mobility, antenna correlation, orbandwidth limited feedback
channel are often simplified oreven omitted in some of previous
works.
This paper presents a new simulator for next
generationMIMO-OFDM-based wireless systems. The proposed simu-lator
runs on top of Wireless Mobile SIMulation (WM-SIM)platform [4],
oriented to model and simulate complex wire-less systems. WM-SIM is
an optimized data flow orientedplatform, based on C++, which is
proved to outperformSimulink, VSS, and Ptolemy II (as discussed
later on). Thesimulator includes the main functionalities carried
out at thePhysical (PHY) and Medium Access Control (MAC) layersof a
wireless MIMO-OFDM system. Besides, the proposed
mailto:[email protected]
-
2 EURASIP Journal on Wireless Communications and Networking
Downlink
Con
trol
plan
e(c
ross
-lay
erad
apta
tion
)
Con
trol
plan
e(m
eas.
and
con
fig.
)
Return
Mobile channel
QoS statistics
Base station User equipments
MAC MAC
PHY PHY
Use
rpl
ane
Use
rpl
ane
Figure 1: Simplified Simulator Architecture.
simulator architecture allows having a fully
configurableMIMO-OFDM transmission, including the main next
gen-eration MIMO-OFDM schemes. The simulator has beenused to
evaluate the performance of the LTE technologyaccording to the
parameters configuration defined by the3GPP standard [1]. The
different MIMO-OFDM techniquesincluded in LTE are evaluated under
realistic assumptionssuch as user mobility, antenna correlation, or
bandwidth-limited feedback channel.
The rest of this paper is structured as follows. First,
adescription of the simulator architecture and functionalitiesis
provided in Section 2. Next, Section 3 presents severalsimulation
results for different LTE scenarios and MIMO-OFDM configurations.
Finally, the main conclusions aresummarized in Section 4.
2. Simulator Architecture
The simulator is composed by a number of User Equipments(UEs)
connected to a Base Station (BS) through a frequency-selective
Rayleigh fading channel. The corresponding high-level architecture
is depicted in Figure 1.
BS and UEs functionalities are split into user andcontrol
planes. In general, the control plane includes mostof the BS/UE
intelligence, acting as a decision point toconfigure user plane
functionalities. In the case of the UEs,control plane is also
responsible for performing physicallayer measurements and for
periodically reporting quality-related information to the BS
through a return channel;such information is used by the BS to
perform a cross-layerlink adaptation, as described in next
subsections. The userplane includes a group of functionalities
(configured fromthe control plane) to facilitate the actual data
transmission.
The simulator also includes a module responsible forcollecting
Quality of Service (QoS) statistics. A detaileddescription of the
whole architecture is given in nextsubsections.
2.1. Base Station. The BS is responsible for schedulingthe
transmission turns to different UEs, for maintaininga reliable
radio link between each UE and the BS (bydynamically adapting the
transmission parameters) as well
as for transmitting data packets to the UEs. A detailed
blockdiagram of the BS architecture (downlink) is depicted inFigure
2.
Incoming data packets arrive at the MAC layer followinga
“full-buffer” model; that are, data is always available at
thetransmission queues. Thus, typical performance indicatorslike
spectral efficiency or Bit Error Rate (BER) are
notservice-dependent. User data at the MAC layer are extractedfrom
the queues according to the decisions taken in the cross-layer
(PHY-MAC) scheduling, as explained later.
Physical layer functionalities are grouped into
threesubsystems.
(i) Modulation and Coding: this subsystem is responsiblefor
channel coding, scrambling, and modulationmapping of each transport
block. The requirednumber of instances of this subsystem is N ,
beingN the number of code words. Throughout thispaper, the term
code word will be employed to referan independently coded and
modulated transportblock. These functions are managed from the
controlplane, where the Modulation and Coding Scheme(MCS)
adaptation is performed and the HybridAutomatic Repeat reQuest
(H-ARQ) block processesthe requested retransmissions of corrupted
codewords.
(ii) MIMO Processing: the use of MIMO technologyrequires
additional processing in order to adequatethe data streams to be
transmitted over the multi-ple antennas. Hence, this subsystem is
responsiblefor mapping and precoding the modulation sym-bols
according to a particular configuration, whichdepends on the MIMO
transmission mode (e.g.,transmit diversity, beamforming, or spatial
multi-plexing). This processing is strongly related to
therank/precoding selection carried out at the controlplane (as
described later).
(iii) OFDM Processing: modulation symbols for eachantenna are
mapped to specific resource elementsin the time-frequency resource
grid. Afterwards, thegeneration of the OFDM time-domain signal for
eachof the M antennas is fulfilled.
-
EURASIP Journal on Wireless Communications and Networking 3
Mod
ula
tion
and
codi
ng
MIM
Opr
oces
sin
gO
FDM
proc
essi
ng
Use
rpl
ane
Con
trol
plan
e
Channel coding
Scrambling
Modulationmapping
Layer mapping
MIMOprecoding
Resourceelement mapping
OFDM modem
Codingscheme
(CS)
Modulation
RI
PMI
Resourceallocation
· · ·
MAC
PHY
1 : N
1 : M
Mantennas
Cro
ss-l
ayer
sch
edu
ling
H-A
RQ
Ran
k/pr
ecod
ing
sele
ctio
n
MC
Sad
apta
tion
-Channel quality indicator (CQI)-Rank indicator (RI)-Precoding
matrix indicator (PMI)-H-ARQ ACK/NACK
· · ·
Figure 2: Base Station Architecture (downlink).
2.1.1. Cross-Layer Scheduling. The time variability of
wirelesschannels leads to rapid changes in the short-term
channelcapacity over time. In an OFDM-based technology, thechannel
response associated to each OFDM subcarrier alsovaries along the
frequency domain. Consequently, chan-nel response can be seen as a
time-frequency randomprocess. In a multiuser environment, channel
quality alsovaries independently for different users. Such
variationsin time-frequency channel conditions for each user canbe
exploited by a cross-layer scheduler to increase
systemthroughput.
Figure 3 shows an example of how the instantaneousChannel State
Information (CSI) is used to perform cross-layer scheduling. Upper
3D graph depicts the instantaneousSignal-to-Noise Ratio (SNR)
received by two users. SNRvalues are periodically reported to the
BS via ChannelQuality Indicators (CQIs), which are used by the
schedulerto decide the time-frequency resource allocation for each
UE(depicted in the lower graph).
When OFDM is jointly used with multiple antennas,a fourth
dimension is added to the scenario. Hence, the
multiplexing algorithm has a high degree of flexibility as
itworks on a time-frequency-space-user basis.
The time-variant radio channel and user specific condi-tions
(e.g., speed) imply the need for adapting or switch-ing the MIMO
configuration to be applied to each user.The adaptation of the MIMO
scheme is accomplished bythe cross-layer scheduler, which selects
dynamically theappropriate scheme for each user. The scheduler
decisionsare based on the per-user reported PHY measurements asCQI,
Rank Indicator (RI), and Precoding Matrix Indicator(PMI). Other
considerations such as the terminal speedor Quality of Service
(QoS) requirements (e.g., priorityhandling) can be also taken into
account in the schedulingprocess.
In order to achieve higher system performance, schedul-ing must
be tightly integrated with MCS adaptation, H-ARQprocesses, and
rank/precoding selection. The schedulingdecisions are based on the
reported CQI, PMI, and RI values,as well as pending HARQ
retransmissions, QoS parameters,and UE capabilities. As a result,
the cross-layer schedulerselects the modulation and coding scheme,
the MIMO
-
4 EURASIP Journal on Wireless Communications and Networking
InstantaneousSNR (dB)
0
5
10
15
20
25
Time Freq
uency
Resourcesallocation
User 1User 2
Figure 3: Example of cross-layer scheduling for two users
[2].
configuration to be applied, and the PHY resources for
datatransmission.
2.1.2. Modulation and Coding. Adaptive Modulation andCoding
(AMC) techniques [2] are used to improve datathroughput whereas BER
is kept below a predefined targetvalue (BERT). Modulation and
coding schemes are jointlychanged by the BS to adapt the
transmitted signal to thevarying channel conditions both in time
and, in frequencydomains.
The control plane is responsible for making decisionson MCS
adaptation, whereas the user plane enforces suchdecisions. MCS
adaptation is based on the instantaneousCQI reported from the
terminals. For the sake of simplicity,only the processing chain for
transport channels is describedbelow as the equivalent one for
control channel is usuallysimilar except for the channel coding
scheme applied.
User plane processing includes the following
functional-ities:
(i) Channel Coding: the channel coding procedure isbased on LTE
specifications [6]. In the first place, aCyclic Redundancy Check
(CRC) is appended to eachtransport block coming from the MAC layer.
Theresulting bit sequence is segmented, if needed, and anadditional
CRC is added to each resulting segment.Then, a turbo coding with
mother coding rate 1/3is applied to each segment for transport
channels.On the other hand, a tail biting convolutionalcoding with
mother rate 1/3 is applied to controlchannels. Finally, a rate
matching process is appliedto each coded segment to match the final
coding rateindicated by the MCS information. Rate-matchedcoded
sequences are concatenated to form a codeword.
(ii) Scrambling: the purpose of this block is to ensurethat the
receiver-side decoding can fully utilize
the processing gain provided by the channel coding.Scrambling
operation consists of an exclusive-oroperation between the code
word and a bit-levelscrambling sequence.
(iii) Modulation Mapping: scrambled code words aremapped onto
the modulation scheme selected fromthe control plane, thus
generating a sequence ofcomplex modulated symbols.
Hybrid-ARQ (Automatic Repeat-reQuest) has become anessential
physical layer feature in mobile communicationsystems like WiMAX or
LTE [7, 8]. H-ARQ allows retrans-missions of a packet at the
MAC/PHY level, with significantadvantages like delay reduction
and/or increased capacity.H-ARQ may be jointly used with Chase
Combining (CC)and Incremental Redundancy (IR) features [9]. When
ChaseCombining is applied, an erroneous code word is stored at
thereceiver in order to be combined with other retransmittedcode
words. Incremental Redundancy is based on the trans-mission of a
punctured version of the original code word.If the decoding fails,
additional redundancy information istransmitted thus having a
different puncturing scheme andhence, facilitating the decoding.
The process may be repeateduntil either a successful decoding or a
maximum numberof retransmissions is reached. H-ARQ procedures are
tightlyintegrated with the cross-layer scheduler, which takes
intoaccount pending retransmissions (from ACK/NACK reports)and
associated redundancy versions.
2.1.3. TransmitMIMOProcessing. The MIMO-OFDM trans-mission
schemes currently being incorporated to next gener-ation wireless
standards have been included in the simulatorin order to provide a
fully configurable MIMO-OFDMtransmission. The transmit MIMO
processing frameworkmay be configured by the control plane in order
to applydifferent MIMO schemes, including both transmit
diversityand spatial multiplexing.
(I) Transmit Diversity: it is used to overcome the effectsof
fading by transmitting redundant informationfrom different
antennas. In particular, the followingschemes are supported:
(i) space-frequency block codes (SFBC),
(ii) transmit beamforming (BF).
(II) Spatial Multiplexing: it is used to increase the
spectralefficiency by transmitting simultaneously several
datastreams from the multiple transmit antennas. Thefollowing
schemes are supported:
(i) open-loop spatial multiplexing (MUX withoutprecoding),
(ii) precoded spatial multiplexing (PrecodedMUX).
The transmit MIMO processing block takes as input thecomplex
modulation symbols of each code word and gives asa result the
complex values to be transmitted on each antenna
-
EURASIP Journal on Wireless Communications and Networking 5
···
a1
a0
Layer mapping
MIMO precoding
···
b3b2b1b0
(2 codewords)
a1a0
b2b0
b3b1x̃
x x0 x1 x2 x3(4 antennas)
(3 layers)
(2 codewords)
RI = 3
(Indicatesa precoding matrix
from codebook)[x = V̂(PMI) · x̃]PMI
(a) Precoded SM with 2 codewords, 3 layers, and 4antennas
···
a1
a0
Layer mapping
MIMO precoding
(1 codeword)
a1a0x̃
x x0 x1 x2 x3
(1 codeword)
(1 layer)
(4 antennas)
RI = 1
(Indicatesa precoding vectorfrom codebook)[x = V̂(PMI) · x̃]
PMI
Complex modulationsymbols
(b) Transmit Beamforming with 4 antennas
Figure 4: Transmit MIMO processing with different configurations
for (a) precoded spatial multiplexing and (b) Transmit
beamforming.
Con
trol
plan
e
Use
rpl
ane
H-A
RQ
(AC
K/N
AC
K)
Ran
k/pr
ecod
ing
esti
mat
ion
SIN
Res
tim
atio
n
Con
trol
sign
allin
g(f
rom
deco
ded
CC
H)
OFD
Mpr
oces
sin
gM
IMO
proc
essi
ng
Dem
odu
lati
onan
dde
codi
ng
Meas. Config. MAC
PHY
1 : N
1 : M
Codingscheme
(CS)
Modulation
RI
PM/RI
Resourceallocation
CSI
channelestimation
OFDM modem
Channeldecoding
De-scrambling
ModulationDe-mapping
LayerDe-mapping
Resource elementDe-mapping
MIMO detection
Mantennas
-Channel quality indicator (CQI)-Rank indicator (RI)-Precoding
matrix indicator (PMI)-H-ARQ ACK/NACK
Figure 5: User Equipment Architecture (downlink).
-
6 EURASIP Journal on Wireless Communications and Networking
Frequency
Time
Pilot symbol
Not used
Time slot #1
Antenna #2
Antenna #1
· · ·
· · ·
l = 5
l = 1
Figure 6: Example of 2D pilot arrangement with
2-antennatransmission.
port. As shown in Figure 2, the complete transmit MIMOprocessing
is separated into two functionalities, layer mappingand MIMO
precoding, which are defined in a different waydepending on the
MIMO scheme to be applied.
(1) Layer Mapping: its purpose is to map the modulationsymbols
of different code words (independentlycoded transport blocks) onto
the spatially multi-plexed streams (layers) that will be
transmitted. Thenumber of code words must be less than or equal
tothe number of layers, which is indicated by the RI.Symbols from
the same code word can be mappedto one or several layers, as shown
in Figure 4. Thenumber of layers may be dynamically adapted to
therank of the channel matrix in spatial multiplexingschemes, thus
making possible to perform rank adap-tation. The cross-layer
scheduler selects the numberof layers of the transmission, which is
forwardedto the layer mapping functionality by means of theRI.
Although the concept of layer does not applyto transmit diversity
schemes, the same mapping isassumed with some distinctions: BF is
considered asa special case of spatial multiplexing with
one-layertransmission, whereas in SFBC the number of layersis
always equal to the number of antennas.
(2) MIMO Precoding: one modulation symbol isextracted from each
layer, being the resulting vectorprocessed according to the applied
MIMO scheme.In case of closed-loop schemes (i.e., BF and
precodedSM) the precoding matrix indicated by PMI isapplied. In the
open-loop SM scheme, the PMIis set to the identity matrix (i.e., no
precoding isapplied). If SFBC is selected (with a reserved valuein
PMI), the coding and mapping procedure definedby SFBC [10] is
performed consequently. The outputof MIMO precoding is an M-size
vector of complexvalues to be mapped onto the M antennas.
An example of MIMO processing is depicted in Figure 4 withtwo
different configurations for three-layer spatial multiplex-ing and
beamforming. The particular configuration of spatialmultiplexing
depicted in Figure 4(a) corresponds to 2 codewords that are
transmitted over 3 spatially multiplexed layersand 4 transmit
antennas. The precoding stage performsthe conversion from 3 layers
to the 4 antennas of thebase station by means of the precoding
matrix, which isselected from the corresponding indexed position
(PMI)within a codebook. On the other hand, Figure 4(b) showsthe
transmit beamforming configuration corresponding to4 transmit
antennas. In this case, only one code wordand one layer are allowed
since no spatial multiplexingwill be performed (i.e., the layer
mapping functionalityis transparent). Therefore, the precoding
matrix becomesa precoding vector, also called beamformer, which is
alsodetermined by the corresponding PMI value.
The configurable MIMO processing framework allowsfor selecting
dynamically the MIMO scheme to be applied todifferent physical
channels. A MIMO scheme is completelydetermined (including
precoding details if necessary) bythe pair of values {RI, PMI}.
Thus, the complete MIMOprocessing is configured by the control
plane with properRI and PMI values for each of the physical
channels beingprocessed.
The adaptation of the MIMO scheme is accomplishedby the
cross-layer scheduler, which selects dynamically theappropriate
scheme for each user. Once the schedulingfunctionality has made the
decision, the different physicaldata channels are processed by the
MIMO framework. Asdetailed above, this framework allows for
changing thesettings of a certain MIMO configuration (e.g., number
oflayers or precoding matrix), as well as switching to a
differentMIMO scheme by simply updating the RI and PMI values.
2.1.4. Transmit OFDM Processing. The generation of theOFDM
signal to be transmitted towards the radio interfacerequires the
following processing blocks.
(i) Resource Element Mapping: resource allocation infor-mation
from the control plane is used to mapmodulation symbols to specific
resource elements inthe time-frequency resource grid for each
antenna.
(ii) OFDM Transmission Modem: once a subframe iscompleted, OFDM
symbols must be sent orderly tothe receiver. In the first place,
OFDM symbols areconverted into the time domain by means of
anM-point Inverse Fast Fourier Transform (M-IFFT).Afterwards, a
cyclic prefix (CP) is added to eachOFDM symbol. Finally, the power
of the transmittedsignal is normalized.
2.2. User Equipment. At the receiver side, UEs receive dataflows
through their multiple antennas. A complex physicallayer processing
is then required until data packets reach theMAC layer. Figure 5
shows the downlink UE architecture,which is divided into a control
plane and a user plane.
-
EURASIP Journal on Wireless Communications and Networking 7
0
100
200
300
400
500
600
700
800
900
1000
Exe
cuti
onti
me
(s)
512 carriers 1024 carriers 2048 carriers
WM-SIMVSS
SimulinkPtolemy II
Figure 7: Execution time consumption for different
simulationplatforms.
Control plane in the UE plays two roles.
(i) Physical Layer Measurements and Reporting: a set
ofindicators are periodically reported to the BS so thatlink
adaptation can be applied at the transmitterside. In particular,
the UE shall estimate the Signalto Interference-plus-Noise Ratio
(SINR), suggest theBS to use a certain rank (RI) and precoding
matrix(PMI) for a given transmit MIMO configuration,and request
potential retransmissions via H-ARQACK/NACK messages.
(ii) Physical Layer Configuration: control channels (CCH)sent
from the BS carry essential control informa-tion for user plane
configuration. Therefore, CCHdecoding (whose details are omitted
for the sakeof simplicity) is the first task to be fulfilled
beforeprocessing user data. Concretely, control channelsinclude
scheduling assignments (i.e., resource allo-cation, modulation and
coding scheme) as well asMIMO-related information like the RI and
PMIapplied by the BS. All this control information isused to
configure the different user plane functions,as depicted in Figure
5.
User plane subsystems are analogous to the ones listed forthe
base station, although specific processing is required atthe
receiver side.
(i) OFDM Processing: received time-domain symbolsthrough each of
the M antennas must be syn-chronized and processed to form a
complete time-frequency resource grid.
(ii) MIMO Processing: this subsystem is in charge ofdetecting
the different data streams (or layers) thathave been transmitted
from the BS. This task requirescertain information from the control
plane, likethe estimated CSI and the RI/PMI used by theBS (decoded
from control channels). In MIMO-OFDM systems, CSI is essential at
the receiver in
order to coherently detect the received signal and toperform
diversity combining or spatial interferencesuppression.
(iii) Demodulation and Decoding: this subsystem isresponsible
for the channel decoding, descramblingand modulation demapping of
each of the N datastreams. The configuration of these functions is
per-formed from the MCS, which is previously decodedfrom control
channels.
2.2.1. Receive OFDM Processing. Each UE must processOFDM signals
from the different receive antennas. This pro-cessing is
differentiated between control plane and user planefunctionalities.
Processing at the user plane is performed bythe two following
blocks.
(i) OFDM Modem: this block is dual to the OFDMtransmission modem
at the BS; that is, cyclic prefixis removed from the received OFDM
symbol in thetime domain; then, an N-FFT is applied to convertOFDM
symbols into the frequency domain. It alsorecovers the reference
pilots required by the channelestimation block.
(ii) Resource Element Demapping: once all the OFDMsymbols
corresponding to a subframe are received,the complete subframe is
processed in a dual manneras done at the transmitter side. That is,
informationfrom the control plane is used to extract the mod-ulated
symbols from their corresponding resourceelements.
At the control plane, several estimates and measurements
arederived upon reception of the OFDM signal. Specifically,
thefollowing functionalities are included.
(i) Channel Estimation: Channel estimation is based onpilot
symbol arrangement within the 2D OFDMresource grid, where pilots
are spread along both timeand frequency domains (see Figure 6).
With this pilotarrangement, an efficient channel estimation
methodmay apply a 2D time-frequency interpolation inorder to
estimate the channel frequency response atall subcarriers within
the complete subframe timeinterval (i.e., for the entire OFDM
resource grid).Channel estimate (Ĥ) is employed with
differentpurposes both in user and control planes at thereceiver.
On the one hand, the channel estimate isprovided to the MIMO
detection block in the userplane to recover the transmitted complex
symbolsfrom the received signals. On the other hand, thechannel
estimate is used to derive several channelmeasurements that will be
reported back to thebase station for link adaptation purposes.
Thesemeasurements include the channel quality, rank, andprecoding
indicators (namely CQI, RI, and PMI).
(ii) SINR Estimation: effective (or postdetected) SINRis
estimated for each frequency subband, given acertain MIMO
transmission mode. The postdetectedSINR depends strongly on the
MIMO scheme being
-
8 EURASIP Journal on Wireless Communications and Networking
Frequency
Tim
e
1200 data subcarriers (20 MHz)
12Subcarriers
· · ·
· · ····
···
TT
I(1
4O
FDM
sym
bols
) Slot
#1Sl
ot#2
PRB #1
PRB #1
PRB #2
PRB #2
PRB#100
PRB#100
Pilot symbol
Not used (reserved for pilots from other antennas)
Minimumallocable unit
Figure 8: LTE physical resources structure.
10−5
10−4
10−3
10−2
10−1
100
BE
R
0 5 10 15 20 25 30
SNR (dB)
MUX w/o precodingPrecoded MUX
SFBCBF
Figure 9: Average BER for different MIMO-OFDM techniques witha
fixed spectral efficiency of 4 bits/s/Hz and low spatial
correlation(v = 3 km/h, ρ = 0.3).
applied and it is derived from the channel estimateĤ. Note that
the MIMO link can be seen as anequivalent end-to-end SISO channel
with an effectiveSNR which depends on the applied MIMO scheme.The
effective SINR is then converted into a quantizedCQI value for
reporting the channel quality back tothe base station through the
uplink control channels.In case of spatial multiplexing, a
different CQI valueis derived for each layer of the
transmission.
(iii) Rank/Precoding Estimation: the suitable rank andpreferred
precoding matrix are derived from thechannel estimate Ĥ for each
frequency subband, sothat RI and PMI values can be reported back to
thebase station for adapting the MIMO transmission.
2.2.2. Receive MIMO Processing. The receive MIMO process-ing
block is in charge of recovering the complex modula-tion symbols
that were transmitted through the multipleantennas. The per-antenna
received complex symbols aretaken as inputs to provide, as a
result, a detected version ofthe complex modulation symbols that
were transmitted foreach code word. Following a similar approach to
that of thetransmitter, the receive MIMO processing is separated
intotwo functionalities: MIMO detection and layer demapping.These
functionalities are defined according to the MIMOscheme applied at
the transmitter, provided that severalschemes are supported in the
simulator (see Section 2.1.3).
The applied MIMO scheme is decided at the base station,which
takes into account the information reported by theUE (i.e.,
suitable rank and preferred precoding matrix).However, the final
scheduling decision may differ withrespect to the reported values
and, therefore, the UE mustbe informed about the applied MIMO
configuration. Hence,the pair of values RI and PMI, which
explicitly determinethe applied MIMO scheme, are sent to the UE via
thecorresponding downlink control channel. At the receiverside, RI
and PMI are extracted from the control channel,which is decoded and
conveniently processed by the controlplane. Finally, RI and PMI are
used to configure the MIMOdetection and layer demapping
functionalities, so that userdata can be properly recovered.
-
EURASIP Journal on Wireless Communications and Networking 9
10−5
10−4
10−3
10−2B
ER
0 5 10 15 20 25 30 35
SNR (dB)
BFMUX w/o precoding
Precoded MUXSFBC
BERT
ASE
(bit
s/s/
Hz)
0 5 10 15 20 25 30 35
SNR (dB)
BFMUX w/o precoding
Precoded MUXSFBC
0
2
4
6
8
10
12
Figure 10: Average BER and ASE with adaptive modulation and BERT
= 10−3 (v = 3 km/h, ρ = 0.3).
(i) MIMO Detection: the objective of the detectionprocess is to
provide the complex modulation sym-bols that were transmitted onto
the different layers.Different detection strategies are implemented
sincethe detection problem is associated to the MIMOtechnique
applied at the transmitter. Linear detectionand low-complexity
algorithms based on successiveinterference cancellation [11] are
supported by thesimulator. In case of SFBC transmission, the
linearprocessing defined in [10] is carried out in orderto obtain
the transmitted complex modulation sym-bols. When precoding-based
schemes are applied,the effective channel matrix is needed in order
toperform the detection. The effective channel Heff =Ĥ∗V is
derived from the channel estimate Ĥ and theprecoding matrix V
indicated by PMI.
(ii) Layer Demapping: it performs exactly the inversefunction of
the layer mapping, described inSection 2.1.3. In this case, the
complex symbolstransmitted over the different layers of
thetransmission are de-mapped into the correspondingcode words to
be subsequently decoded (as describedin the next section).
2.2.3. Demodulation and Decoding. At the receiver side,incoming
modulated symbols are converted to bit sequencesand processed in
order to recover the original transportblocks. The UE is aware of
the instantaneous MCS appliedto the data from the signalling
information contained in thecorresponding control channels. The
following processingblocks are included.
(i) Modulation Demapping: the sequence of receivedcomplex
symbols is demapped according to the MCSinformation provided by the
control plane.
(ii) Descrambling: the scrambling process carried out atthe
receiver is undone to recover the descrambledcode word.
(iii) Channel Decoding: the received code word has tobe
processed to reverse the rate-matching processbefore being decoded.
Once the decoding is complete,the transport block CRC integrity is
checked. Ifsegmentation took place at the transmitter, the CRCof
each individual segment is checked before.
If data integrity is preserved, an ACK message is sent tothe
transmitter through uplink control channels. Otherwise,a NACK
message is sent to request a retransmission. IfCC and/or IR
features are applied, the erroneous codeword is stored at the
receiver to be combined (CC) oradded (IR) to forthcoming code
words, as described inSection 2.1.2.
2.3. Mobile Channel. The mobile channel subsystem
allowssimulating a MIMO transmission through the radio link.
Theeffects of antenna spatial correlation and
frequency-selectiveRayleigh fading are included. Specifically, the
standardspatially correlated Rayleigh-faded multiantenna
channelmodel [12] is considered. Channel gain is modeled by a
2×2complex matrix H, so that the entries H = (hi j) denote
thechannel gain between the jth transmit and the ith
receiveantenna. The multitap channel delay profile is fully
con-figurable. Assuming the well-known Kronecker
correlationstructure [12], the channel matrix can be decomposed asH
= R1/2rx · G · R1/2tx , where the entries of G are independentand
identically distributed (i.i.d.) Gaussian random variables(RVs)
with zero mean and unit variance. We assume the sameantenna
correlation factor (ρ) for transmit and for receive
-
10 EURASIP Journal on Wireless Communications and Networking
5 10 15 20 25 30 35 40
Mea
nde
lay
(s)
Load (Mbps)
0
0.02
0.04
0.06
0.08
0.1
BFPrecoded MUX
MUX w/o precodingSFBC
Mean SNR = 10 dB
(a)
70 80 90 100 110 120 130 140
Mea
nde
lay
(s)
Load (Mbps)
0
0.02
0.04
0.06
0.08
0.1
BFPrecoded MUX
MUX w/o precodingSFBC
Mean SNR = 25 dB
(b)
Figure 11: Mean transmission delay for different MIMO schemes
when using adaptive modulation subject to restriction BERT = 10−3
in apedestrian channel (v = 3 km/h, ρ = 0.3) with different average
SNRs.
antennas, thus being the correlation matrices Rtx and
Rrxidentical and given by
Rtx = Rrx =⎛⎝ 1 ρρ∗ 1
⎞⎠. (1)
The effect of Additive White Gaussian Noise (AWGN) at thereceive
antennas is also included. Assuming the frequencydomain baseband
model, the received signal vector can beexpressed as y = H · x + n,
where x is the transmittedsignal vector and n is the channel noise
vector, whose entriesare i.i.d. Gaussian RVs with zero mean and
variance σ2n . Asthe transmitted signal is power normalized, the
average SNRdetermines the average noise power. Thus, we define
theaverage SNR γ in terms of the transmit power constraint andthe
noise power as γ = 1/σ2n [13].
The simulator includes the ability to run over a timedomain
channel model (with higher accuracy) or over anequivalent frequency
domain channel model (to minimizecomputational costs). If the later
is selected, OFDM modemsare disabled, being OFDM symbols merely
affected by chan-nel frequency response. However, it should be
noted thatperformance results over an equivalent frequency
domainchannel are only valid under the premise that transmissionis
free of Intersymbol Interference (ISI) and IntercarrierInterference
(ICI).
2.4. QoS Statistics. A QoS statistics block allows assessingthe
impact of PHY layer impairments on upper layers’QoS by collecting
performance statistics from both sides ofthe communication (as
shown in Figure 1). This block isaware of the status of the BS and
all UEs in the simulationscenario (queue occupancy at BS,
transmitted/received bits
sequences, etc.). This global knowledge allows obtaining
QoSindicators like BER, BLock Error Rate (BLER), transmissiondelay,
throughput, as well as occupancy and loss rate in thetransmission
queues.
3. Simulation Results
The presented simulator allows for evaluating the mainphysical
and MAC layer functionalities of a MIMO-OFDMA-based system. The
performance of different adaptive trans-mission techniques and MIMO
schemes may be evaluated ina complete downlink system that takes
into account most ofthe impairments of a realistic scenario.
As aforementioned, the proposed simulator is proved tooutperform
Simulink, VSS and Ptolemy II. Figure 7 showsa performance
comparison among these platforms for asimplified OFDM scenario (see
[4] for further details).
This simulator has been used to support several researchworks in
the context of adaptive OFDMA, and MIMO basedsystems. In [11],
different low-complexity MIMO detectionalgorithms are evaluated in
a realistic LTE downlink scenarioand a performance-complexity
tradeoff is established. Anadaptive MIMO transmission is proposed
in [14] for anOFDMA-based cellular system under practical
limitationssuch as imperfect CSI at the transmitter due to user
mobility.
Along this section, simulation results are shown for atypical
configuration of the 3GPP LTE technology [1], whichis summarized in
the next subsection.
3.1. LTE Configuration. LTE radio resources are structuredin
slots of length Tslot = 0.5 millisecond. The TransmissionTime
Interval (TTI) is set to 1 millisecond, thus containing 2
-
EURASIP Journal on Wireless Communications and Networking 11
10−5
10−4
10−3
10−2B
ER
0 5 10 15 20 25 30 35
SNR (dB)
BF (10 kmh)BF (20 kmh)BF (30 kmh)
BF (40 kmh)BF (60 kmh)BF (80 kmh)
BERT
10−5
10−4
10−3
10−2
BE
R
0 5 10 15 20 25 30
SNR (dB)
SFBC (60 kmh)SFBC (70 kmh)SFBC (80 kmh)
Precoded MUX (10 kmh)Precoded MUX (20 kmh)Precoded MUX (30
kmh)
BERT
Figure 12: Evaluation of the maximum admissible user speed for
different MIMO schemes under adaptive modulation.
10−5
10−4
10−3
10−2
BE
R
0 5 10 15 20 25 30 35
SNR (dB)
BF-idealPrecoded MUX-ideal
BF-codebookPrecoded MUX-codebook
BERT
(a)
ASE
(bit
s/s/
Hz)
0 5 10 15 20 25 30 35
SNR (dB)
BF-idealPrecoded MUX-ideal
BF-codebookPrecoded MUX-codebook
0
2
4
6
8
10
12
(b)
Figure 13: Performance degradation due to codebook based
precoding.
slots. Each slot can be seen as a time-frequency resource
gridcomposed by several OFDM symbols. Each slot is dividedinto a
number of Physical Resource Blocks (PRBs), eachof them consisting
of 12 consecutive subcarriers along 7consecutive OFDM symbols.
Assuming a system bandwidthBW = 20 MHz, a total of 1200 data
subcarriers (i.e., 100PRBs) is available for transmission. The
minimum allocableunit to a user is formed by two consecutive PRBs
along thetime. Although LTE specifications state that the first
OFDMsymbols (from 1 to 3) in a TTI are reserved for controlchannels
(signaling), our simulations only assume referencesignals overhead,
as shown in Figure 8.
A typical value of antenna correlation ρ = 0.3 has
beenconsidered from previous works on the antenna correlationfor a
typical suburban scenario and linear antenna arrays[15]. A 9-tap
typical urban channel delay profile hasbeen considered [16]. The
rest of simulation parameters issummarized in Table 1.
3.2. Performance Results. In this section, several
simulationresults are shown in order to provide a comparative
analysisof the different MIMO-OFDM techniques in the LTEdownlink
scenario. MIMO-OFDM techniques (SFBC, BF,MUX without precoding, and
precoded MUX) are evaluated
-
12 EURASIP Journal on Wireless Communications and Networking
50
60
70
80
90
100
110
120
130
140
Cel
lTh
rou
ghpu
t(M
bps)
0 5 10 15 20 25 30 35 40 45 50
Number of users
Precoded MUXMUX w/o precoding
BFSFBC
Figure 14: Multiuser diversity gain with Proportional Fair
schedul-ing (average SNR 20 dB).
Table 1: Simulation parameter values.
Parameter Description Value
Nant Number of Antennas 2× 2ρ Antenna Correlation 0.3
— MIMO TransmissionMode
MUX, PrecodedMUX, SFBC, BF
v Terminal Speed 3–80 km/h
γ Average SNR 0–30 dB
— ModulationQPSK, 16QAM,64QAM
TTI Transmission TimeInterval
1 ms
fc Carrier Frequency 2.5 GHz
BW Bandwidth 20 MHz
fs Sampling Frequency 30.72 MHz
Δ f Subcarrier Spacing 15 kHz
FFTsize FFT Size 2048
Nsub Data subcarriers 1200
rsReference signalsoverhead
2/21
CP Cyclic Prefix Length 144 samples
— Source Model Full buffer
t Simulation Length 20 seconds
in terms of average BER, average spectral efficiency (ASE),and
mean transmission delay.
The presented results are organized as follows. First,
theaverage BER is evaluated without any adaptive
modulationmechanism, that is, under a fixed given spectral
efficiency.Afterwards, the performance analysis is focused on
theadaptive modulation mechanism and, thus, the average
BER,spectral efficiency, and mean transmission delay are
provided
for the different MIMO-OFDM schemes. Next, differentimpairments
associated to a realistic scenario are tackled.Specifically, the
maximum admissible user speed under agiven target BER (BERT)
requirement is evaluated for thedifferent MIMO techniques. Finally,
the performance degra-dation due to nonideal (i.e., codebook based)
precoding isaddressed.
Figure 9 shows the average BER results associated to thefour
MIMO-OFDM techniques. In order to provide a faircomparative
analysis, a fixed spectral efficiency of 4 bits/s/Hzis achieved by
employing different modulation order for thedifferent MIMO
techniques; that is, QPSK is employed forspatial multiplexing
schemes whereas 16-QAM is used in caseof transmit diversity (i.e.,
SFBC and BF schemes). Simulationresults show that the transmit
diversity schemes outperform(for the entire range of SNR) the
multiplexing ones in termsof BER. Besides, precoding allows
reducing significantly theinterference among transmitted signals at
the receiver and,therefore, it improves the performance when
applied to theMUX technique.
Figures 10 and 11 show different performance metrics ofthe
different MIMO-OFDM schemes with adaptive modula-tion. An
instantaneous BER restriction is assumed for all thesimulations so
that BER is maintained below a predefinedtarget (BERT = 10−3).
Possible modulation schemes areQPSK, 16-QAM, and 64-QAM, as well as
no transmission(outage). An error-free feedback channel is assumed
andideal precoding based on singular value decomposition(SVD) of
the channel matrix is employed for the BF andprecoded MUX
techniques.
In Figure 10, the average BER and corresponding ASE isdepicted
as a function of the average SNR for the differentMIMO-OFDM
schemes. It can be seen that the average BERremains under the BERT
within the entire SNR range forall the schemes. Regarding the ASE,
different results for thelow and high SNR regime are obtained (see
Figure 10). Inthe low SNR regime, BF provides the highest ASE
comparedto other schemes; however, the achieved ASE for
transmitdiversity techniques (both BF and SFBC) reaches a
satura-tion level (6 bits/s/Hz) as the average SNR increases.
Thissaturation level corresponds to the maximum achievable ASEby
transmitting only one data stream with the maximummodulation order
(64-QAM). On the other hand, the spatialmultiplexing techniques
provide the highest ASE (close to12 bits/s/Hz) in the high SNR
regime due to the simultaneoustransmission of two data streams.
Moreover, a significantASE improvement is shown for the MUX
technique whenprecoding is applied (ASE increases 2 bits/s/Hz when
averageSNR is 20 dB).
Figure 11 illustrates the mean transmission delay asa function
of the system load for the different MIMO-OFDM schemes being
compared. The load point at whichmean delay starts increasing
determines the maximumadmissible system load. Two different
scenarios are con-sidered in Figure 11 with average SNR of 10 dB
(a) and25 dB (b), respectively. The ASE differences between theMIMO
schemes in the low and high SNR regime describedabove are also
observed in Figure 11 in terms of the meantransmission delay. In
the low SNR regime (Figure 11(a)) the
-
EURASIP Journal on Wireless Communications and Networking 13
limit for the admissible load with BF is about 30–35
Mbps,significantly higher than the one achieved with the
otherschemes. In case of high SNR regime (Figure 11(b)), it isshown
that MUX schemes (specially the precoded MUX)clearly outperform the
transmit diversity schemes.
Figure 12 shows how average BER is degraded as userspeed
increases due to the outdated CQI and PMI reports.It is observed
that the maximum user speed that fulfills theBER requirement for BF
is around 60–70 km/h, which issignificantly higher that the one for
precoded MUX scheme(about 20 km/h). However, SFBC is shown to be
the mostrobust technique in high mobility scenarios as the
maximumadmissible user speed is around 80–90 km/h. This is due
tothe fact that, unlike the BF scheme, SFBC is not affected bythe
outdated PMI reports.
Another impairment associated to a realistic sce-nario is the
bandwidth limited feedback channel, whichmakes unfeasible to report
the exact (ideal) precodingmatrix to the base station. Therefore,
codebook-basedprecoding is employed and only an index which
indi-cates a predefined precoding matrix (within the code-book) is
reported. Figure 13 shows the performancedegradation in terms of
average BER (a) and ASE (b)when a codebook-based precoding is
applied insteadof ideal (SVD based) precoding. The two-bit
codebookdefined in LTE [17] is assumed for the simulations. Itis
observed that the performance degradation (both inaverage BER and
ASE) associated to BF is negligible,while the precoded MUX
performance is degraded signifi-cantly.
As described in Section 2.1.1, the scheduling policyapplied at
the MAC layer plays an important role inthe overall system
performance. Figure 14 shows the totalcell throughput for a
different number of users in thecell when applying a Proportional
Fair (PF) schedulingpolicy. Simulation results show the multiuser
diversity gainachieved by PF for different MIMO schemes, which
canbe seen as an effective SNR gain in the radio link. ForSFBC and
BF, the maximum achievable cell throughput isRcell ≈ 91.5 Mbps,
which corresponds to the maximumASE of 6 bits/s/Hz shown in Figure
10, taking into accountthe physical resources structure defined in
Table 1, that is,Rcell ≈ ASE · Nsub · (1 − rs) · fs/(CP + FFTsize).
AlthoughMUX without precoding is the worst scheme for a singleuser
in the cell, both MUX-based schemes achieve thehighest multiuser
diversity gain as the transmission of twosimultaneous streams
provides a higher degree of freedom.Anyhow, precoded MUX
outperforms the previous schemeboth in throughput and BER due to
the reduction of thecochannel interference achieved by means of
precoding.
4. Conclusions
In this paper, we present a simulator for next genera-tion
MIMO-OFDM-based wireless systems. The simulatorincludes the main
functionalities carried out at the physicaland MAC layers of a
wireless MIMO-OFDM system. Furtheron, a detailed description of the
simulator architecture andmain functionalities is provided. With
this architecture, it is
shown that a fully configurable MIMO-OFDM transmissionis
supported, including all the MIMO-OFDM schemes beingincorporated to
next generation wireless standards.
This simulator has been used to evaluate the perfor-mance of the
3GPP LTE technology, whose simulationresults are here presented and
analyzed. Performance resultsare provided for the different
MIMO-OFDM techniquesincluded in LTE under realistic assumptions
such as usermobility or bandwidth limited feedback channel. It has
beenproved that spatial multiplexing techniques provide the
bestspectral efficiency for high SNR although very low
terminalspeeds (up to 20 km/h) are supported to fulfill the
reliabilityrequirements. On the other hand, SFBC is shown to be
themost robust technique in high mobility scenarios as themaximum
admissible user speed is around 80–90 km/h.
Acknowledgments
This work has been partially supported by the Spanish
Gov-ernment (projects TIC2003-07819 and TEC2007-67289),Junta de
Andalucı́a (Proyecto de Excelencia TIC 03226), andAT4wireless.
References
[1] 3GPP TS 36.201, “Long Term Evolution (LTE) physical
layer;General description,” Release 8, V8.3.0, March 2009.
[2] G. Gómez, D. Morales-Jiménez, F. J. López-Martinez, J.
J.Sánchez, and J. T. Entrambasaguas, “Radio-interface
physicallayer,” in Long Term Evolution: 3GPP LTE Radio and
CellularTechnology, chapter 3, pp. 49–98, Auerbach, Boca Ratón,
Fla,USA, April 2009.
[3] C. Mehlfuhrer, M. Wrulich, J. C. Ikuno, D. Bosanska, andM.
Rupp, “Simulating the long term evolution physical layer,”in
Proceedings of 17th European Signal Processing Conference(EUSIPCO
’09), pp. 1471–1478, Glasgow, Scotland, August2009.
[4] J. J. Sánchez, D. Morales-Jiménez, G. Gómez, E.
Martos-Naya, U. Fernández-Plazaola, and J. T.
Entrambasaguas,“WM-SIM: a platform for design and simulation of
wirelessmobile systems,” in Proceedings of the 2nd ACM Workshop
onPerformance Monitoring and Measurement of HeterogeneousWireless
and Wired Networks (PM2HW2N ’07), pp. 124–127,Chania, Greece,
October 2007.
[5] D. Martı́n-Sacristán, J. F. Monserrat, J.
Cabrejas-Peñuelas,D. Calabuig, S. Garrigas, and N. Cardona, “On
the waytowards fourth-generation mobile: 3GPP LTE and
LTE-advanced,” EURASIP Journal on Wireless Communications
andNetworking, vol. 2009, Article ID 354089, 10 pages, 2009.
[6] J. Lee, J.-K. Han, and J. Zhang, “MIMO technologies in3GPP
LTE and LTE-advanced,” EURASIP Journal on WirelessCommunications
andNetworking, vol. 2009, Article ID 302092,10 pages, 2009.
[7] J. J. Sánchez, D. Morales-Jiménez, G. Gómez, and J.
T.Enbrambasaguas, “Physical layer performance of long termevolution
cellular technology,” in Proceedings of the 16th ISTMobile and
Wireless Communications Summit, July 2007.
[8] M. Ergen, Mobile Broadband Including WiMAX and LTE,Springer
Science+Business Media LLC, 2009.
[9] P. Frenger, S. Parkvall, and E. Dahlman, “Performance
com-parison of HARQ with chase combining and incremental
-
14 EURASIP Journal on Wireless Communications and Networking
redundancy for HSDPA,” in Proceedings of the 54th IEEEVehicular
Technology Conference (VTC ’01), vol. 3, pp. 1829–1833, Atlantic
City, NJ, USA, October 2001.
[10] M. J. Dehghani, R. Aravind, S. Jam, and K. M. M.
Prabhu,“Space-frequency block coding in OFDM systems,” in
Pro-ceedings of IEEE Region 10 Conference: Analog and
DigitalTechniques in Electrical Engineering (TENCON ’04), pp.
A543–A546, Chiang Mai, Taiwan, November 2004.
[11] D. Morales-Jiménez, J. F. Paris, and J. T.
Entrambasaguas,“Performance tradeoffs among low-complexity
detectionalgorithms for MIMO-LTE receivers,” International Journal
ofCommunication Systems, vol. 22, no. 7, pp. 885–897, 2009.
[12] D.-S. Shiu, G.J. Foschini, M.J. Gans, and J. M. Kahn,
“Fadingcorrelation and its effect on the capacity of
multielementantenna systems,” IEEE Transactions on Communications,
vol.48, no. 3, pp. 502–513, 2000.
[13] A. Goldsmith, Wireless Communications, Cambridge
Univer-sity Press, New York, NY, USA, 2005.
[14] D. Morales-Jiménez, G. Gómez, J. F. Paris, and J. T.
Entram-basaguas, “Joint adaptive modulation and MIMO transmis-sion
for non-ideal OFDMA cellular systems,” in Proceedingsof the 5th
IEEE Broadband Wireless Access Workshop Colocatedwith IEEE
GLOBECOM, Honolulu, Hawaii, USA, December2009.
[15] K. I. Pedersen, J. Andersen, J. Kermoal, and P. Mogensen,“A
stochastic multiple-input-multiple-output radio channelmodel for
evaluation of space-time coding algorithms,” inProceedings of the
52nd IEEE Vehicular Technology Conference(VTC ’00), vol. 2, pp.
893–897, Boston, Mass, USA, September2000.
[16] 3GPP TS 36.521-1, “User Equipment (UE)
conformancespecification Radio transmission and reception, Part 1:
Con-formance Testing,” Release 8, V8.1.0, March 2009.
[17] 3GPP TS 36.211, “Physical Channels and Modulation,”Release
8, V8.6.0, March 2009.
IntroductionSimulator ArchitectureBase StationCross-Layer
SchedulingModulation and CodingTransmit MIMO ProcessingTransmit
OFDM Processing
User EquipmentReceive OFDM ProcessingReceive MIMO
ProcessingDemodulation and Decoding
Mobile ChannelQoS Statistics
Simulation ResultsLTE ConfigurationPerformance Results
ConclusionsAcknowledgmentsReferences