Mehlführer et al.

EURASIP Journal on Advances in Signal Processing 2011, 2011:29

http://asp.eurasipjournals.com/content/2011/1/29

RESEARCH Open Access

The Vienna LTE simulators - Enabling

reproducibility in wireless communications
research
Christian Mehlführer*, Josep Colom Ikuno, Michal Šimko, Stefan Schwarz, Martin Wrulich and Markus Rupp

Abstract
In this article, we introduce MATLAB-based link and system level simulation environments for UMTS Long-Term
Evolution (LTE). The source codes of both simulators are available under an academic non-commercial use license,
allowing researchers full access to standard-compliant simulation environments. Owing to the open source
availability, the simulators enable reproducible research in wireless communications and comparison of novel
algorithms. In this study, we explain how link and system level simulations are connected and show how the link
level simulator serves as a reference to design the system level simulator. We compare the accuracy of the PHY
modeling at system level by means of simulations performed both with bit-accurate link level simulations and
PHY-model-based system level simulations. We highlight some of the currently most interesting research questions
for LTE, and explain by some research examples how our simulators can be applied.
Keywords: LTE, MIMO, link level, system level, simulation, reproducible research

1. Introduction reviewers of articles) to reproduce the results presented.

Reproducibility is one of the pillars of scientific research. Unfortunately, researcher’s reality does not resemble
Although reproducibility has a long tradition in most this ideal situation, a circumstance that has recently
nature sciences and theoretical sciences, such as mathe- been quite openly complained about [3].
matics, it is only recently that reproducible research has In the past few years, several researchers have started
become more and more important in the field of signal to build up online resource databases in which simula-
processing [1,2]. In contrast to results in fields of purely tion code and data are provided, see for example [4,5].
theoretical sciences, results of signal processing research However, it is still not a common practice in signal pro-
articles can be reproduced only if a comprehensive cessing research. We are furthermore convinced that
description of the investigated algorithms (including the reproducibility should also play an important role in the
setting of all necessary parameters), as well as eventually review process of an article. Although thorough check-
required input data are fully available. Owing to the lack ing is very possibly impractical, it would make the pre-
of space, a fully comprehensive description of the algo- sented studies more transparent to the review process.
rithm is often omitted in research articles. Even if an Reproducibility becomes even more important when the
algorithm is explained in detail, for instance, by a systems that are simulated become more and more
pseudo code, initialization values are often not fully complex, as it is the case in the evaluation of wireless
defined. Moreover, it is often not possible to include in communication systems. When algorithms for wireless
an article all the necessary resources, such as data, systems are evaluated, authors often claim to use a stan-
which were processed by the presented algorithms. Ide- dard-compliant transmission system and simply make
ally, all resources, including source code of the pre- reference to the corresponding technical specification.
sented algorithms, should be made available for Since technical specifications are usually extensive,
download to enable other researchers (and also including a cornucopia of options, it is not always clear
which parts of a specification were actually implemented
* Correspondence: [email protected] and which parts were omitted for the sake of simplicity
Institute of Telecommunications, Vienna University of Technology, Austria

© 2011 Mehlführer et al; licensee Springer. This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 2 of 14
http://asp.eurasipjournals.com/content/2011/1/29

reasons. The situation of trying to reproduce someone dramatically) [14], modeling of channel encoding and
else’s results to compare them to one’s own algorithm decoding [15], physical-layer modeling crucial, for sys-
but not being able to do so (or only after extensive tem level simulations [16], and the like are typically ana-
effort to discover the unreported details of the actual lyzed on link level. Although MIMO broadcast channels
implementation) is familiar to most researchers. With- have been investigated quite extensively over the past
out access to the details of the implementation, includ- few years [17,18], there are still a lot of open questions
ing all assumptions, comparisons of algorithms, that need to be resolved, both in theory and in practical
developed by different researchers, are very difficult, if implementation. For example, LTE offers the flexibility
not impossible to carry out. A way out of this dilemma to adjust many transmission parameters, but it is not
is offered by a publicly available simulation environ- clear up to now how to exploit the available Degrees of
ment. In this study, we present such an open-source Freedom (DoF) to achieve the optimum performance.
simulation environment that supports link and system Some recent theoretical results point out how to pro-
level simulations of the Universal Mobile Telecommuni- ceed in this matter [18,19], but practical results for LTE
cations System (UMTS) Long-Term Evolution (LTE), are still missing.
specifically designed to support reproducibility. The 2) System level simulations focus more on network-
development and publishing of this LTE simulation related issues, such as resource allocation and schedul-
environment is based on our previous very good experi- ing [20,21], multi-user handling, mobility management,
ence with a WiMAX physical layer simulator [6]. admission control [22], interference management
Furthermore, such simulators can be used as a refer- [23,24], and network planning optimization [25,26].
ence for validation of algorithms, for example, when Furthermore, in a multi-user oriented system, such as
designing transmitter or receiver chips [7]. We also have LTE, it is not directly clear which figures of merit
used our simulators for generating LTE signals that are should be used to assess the performance of the system.
required to include realistic signals in related research The classical measures of (un)coded Bit Error Ratio
[8], or as a reference for LTE-compliant measurements. (BER), (un)coded BLock Error Ratio (BLER), and
In such cases, the simulator can serve not only as a data throughput are not covering multi-user scenario proper-
pump, but also as a vehicle to evaluate the received data. ties. More comprehensive measures of the LTE perfor-
LTE, the current evolutionary step in the third Gen- mance are, for example, fairness, multi-user diversity, or
eration Partnership Project (3GPP) roadmap for future DoF [27]. However, these theoretical concepts have to
wireless cellular systems, was introduced in 3GPP be mapped to performance values that can be evaluated
Release 8 [9]. Besides the definition of the novel physical by means of simulations [28,29].
layer, LTE also contains many other remarkable innova- Around the world, many research facilities and ven-
tions. Most notable are (i) the redevelopment of the sys- dors are investigating the above mentioned aspects of
tem architecture, now called System Architecture LTE. For that purpose, commercially available simula-
Evolution (SAE), (ii) the definition of network self-orga- tors applied in industry [30-32], as well simulators
nization, and (iii) the introduction of home base-sta- applied in academia [33], have been developed. Also,
tions. The main reasons for these profound changes in probably all major equipment vendors have implemen-
the Radio Access Network (RAN) system design are to ted their own, proprietary simulators. Regardless of the
provide higher spectral efficiency, lower delay (latency), simulation tools being commercial/noncommercial, the
and more multi-user flexibility than the currently development framework (C, C++, MATLAB, WM-SIM
deployed networks. [33],...), or their claimed performance/flexibility, one fact
In the development and standardization of LTE, as is shared by all of the simulators. Their closed imple-
well as in the implementation process of equipment mentation disables access to implementation details and
manufacturers, simulations are necessary to test and thus to any assumption that may have been included.
optimize algorithms and procedures. This has to be car- As such, the reliability of the results relies purely on the
ried out on the physical layer (link level) and in the net- faith of a proper implementation. Independent valida-
work (system level) context: tion of results in such closed simulation environments is
1) Link level simulations allow for the investigation of not easy, very time-consuming, and often not feasible.
channel estimation, tracking, and prediction algorithms, Since the results were obtained with closed tools, simply
as well as synchronization algorithms [10,11]; Multiple- repeating the same experiment is a daunting task.
Input Multiple-Output (MIMO) gains; Adaptive Modu- Transparency not only in the results, but also in the
lation and Coding (AMC); and feedback techniques tools employed, thus greatly magnifies the credibility of
[12,13]. Furthermore, receiver structures (typically the results.
neglecting inter-cell interference and impact of schedul- The two simulators [34,35] described in Sections 2
ing, as this increases simulation complexity and runtime and 3 of this article are freely available at our homepage
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 3 of 14
http://asp.eurasipjournals.com/content/2011/1/29

http://www.nt.tuwien.ac.at/ltesimulator/ under an open, signaling is stronger protected than data, by means of

free for non-commercial academic use license, which lower coding rates and/or lower-order modulations, the
facilitates academic research and enables a closer coop- assumption of error-free signaling is in fact quite realis-
eration between different universities and research facil- tic. Equivalently, errors on the signaling channels will
ities. In addition, developed algorithms can be shared only occur when the data channels are already facing
under the same license again, making the comparison of severe performance degradation–a point of operation
algorithms easier, reproducible, and therefore refutable usually not targeted in investigations.
and more credible. To the best of the authors’ knowl- In the downlink, the signaling information passed on
edge, our two simulators are the first to be published in by the transmitter to the receiver contains coding,
the context of LTE, including source code under an aca- HARQ, scheduling, and precoding parameters. In the
demic use license. Thus, the simulators provide oppor- uplink, Channel Quality Indicator (CQI), Precoding
tunities for many institutions to directly apply their Matrix Indicator (PMI), and Rank Indicator (RI) are sig-
ideas and algorithms in the context of LTE. The avail- nalled, which together form the Channel State Informa-
ability of the simulators, together with the possibility to tion (CSI) feedback. All simulation scenarios (see
include links to the utilized simulator version and any Section 2-B) support the feedback of CQI, PMI, and RI,
resources needed furthermore, enables researchers to although it is also possible to set some or all of them to
quickly reproduce published results [2]. fixed values. Such a setting is required for specific simu-
The remainder of this article is organized as follows. lations, such as throughput evaluation of a single Modu-
In Sections 2 and 3, we describe the Vienna LTE Simu- lation and Coding Scheme (MCS).
lators and how they relate to each other. In Section 4, A standard-compliant implementation of the downlink
we provide a validation of the two simulators. Exemp- control channels would not affect the overall structure
lary simulation results are shown in Section 5. Finally, of our simulator and just requires the insertion of the
we conclude the article in Section 6. control channels in the relevant resource elements [36].
On the other hand, non-error-free feedback transmis-
2. The Vienna LTE link level simulator sions would require a physical layer implementation of
In this section, we describe the overall structure of the the LTE uplink, which is currently not in the scope of
Vienna LTE Link Level Simulator, currently (January the simulator. (A first release of the uplink, however, is
2011) released in version 1.6r917. Furthermore, we pre- currently being implemented in the simulator and will
sent the capabilities of the simulator and provide some be released soon.)
examples of its application. 1) Transmitter
The layout of the transmitter is shown in Figure 2, which
A. Structure of the simulator is also a graphical representation of the transmitter
The link level simulator can be divided into three basic
building blocks, namely, transmitter, channel model, and
receiver (see Figure 1). Depending on the type of simula- user feedback
tion, one or several instances of these basic building
blocks are employed. The transmitter and receiver data bits random data bits
HARQ control

blocks are linked by the channel model, which is used allowed generation

to transmit the downlink data, while signaling and

coding
uplink feedback is assumed to be error-free. Since params per-user channel coding
of the data bits
scheduler

modulation
symbol mapping
PDP-based channel or params
Winner+ channel trace
precoding layer mapping, reference/sync
Channel coded/uncoded BER params precoding
model symbols
TX RX block error rate
signaling
throughput RB
OFDM symbol assembly
allocation
CSI feedback, IFFT
Delay
ACK/NACKs
CP insertion
Figure 1 LTE link level simulator overall structure, as
implemented in the Vienna LTE link level simulator. The
signaling transmitted signal
simulator comprises by one or more transmitter blocks, channel
modeling for each link, and receiver blocks. The feedback channel is Figure 2 LTE downlink transmitter implementation in the
implemented as a delayed error-free signaling channel. Vienna LTE link level simulator, as specified in [36-38].
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 4 of 14
http://asp.eurasipjournals.com/content/2011/1/29

description defined in the TS36’ standard series [36-38].

received signal signaling
Based on User Equipment (UE) feedback values, a sche-
duling algorithm assigns Resource Blocks (RBs) to UEs CP removal
and sets an appropriate MCS (coding rates between FFT
0.076 and 0.926 with 4, 16, or 64-QAM modulation [38]),
the MIMO transmission mode (Transmit Diversity time-frequency
resource block grid
(TxD), Open Loop Spatial Multiplexing (OLSM), or RB allocation
Closed Loop Spatial Multiplexing (CLSM)), and the pre- CQI/PMI/RI
channel resource block
coding/number of spatial layers for all served users. Such feedback
estimation disassembling
calculation
a channel adaptive scheduling allows for the exploitation
of frequency diversity, time diversity, spatial diversity, precoding
MIMO RX and OFDM detection
and multi-user diversity.
coding
Given the number of available DoF, the specific imple- channel decoding
params
mentation of the scheduler algorithm has a large impact
user decoded BER
on the system performance and has been a hot topic in feedback data bits BLER
throughput
research [39-41]. In Section 5-B, we provide perfor-
Figure 3 LTE downlink receiver structure, as implemented in
mance evaluations of several schedulers. the Vienna LTE link level simulator.
2) Channel model
The Vienna LTE Link Level Simulator supports block-
and fast-fading channels. In the block-fading case, the specifies three feedback indicators for that purpose:
channel is constant during the duration of one subframe CQI, RI, and PMI [36]. The CQI is employed to choose
(1 ms). In the fast-fading case, time-correlated channel the appropriate MCS, such as to achieve a predefined
impulse responses are generated for each sample of the target BLER, whereas the RI and the PMI are utilized
transmit signal. Currently (January 2011), the simulator for MIMO pre-processing. Specifically, the RI informs
supports the following channel models: the eNodeB about the preferred number of parallel spa-
tial data streams, while the PMI signals the preferred
1) Additive White Gaussian Noise (AWGN); precoder that is stemming from a finite code book as
2) Flat Rayleigh fading; specified in [36]. Very similar feedback values are also
3) Power Delay Profile-based channel models, such employed in other systems such as WiMAX and WiFi.
as ITU Pedestrian B, or ITU Vehicular A [42]; The simulator provides algorithms that utilize the esti-
4) Winner Phase II+ [43] mated channel coefficients to evaluate these feedback
indicators [13]. Researchers and engineers working on
The most sophisticated of these channel models is the feedback algorithms can implement other algorithms
Winner Phase II+ model. It is an evolution of the 3GPP using the provided feedback functions as a starting
spatial channel model, and introduces additional fea- point to define their own functions.
tures, such as support for arbitrary 3D antenna patterns. Given this receiver structure, the simulator allows the
3) Receiver investigation of various aspects, such as frequency syn-
Figure 3 shows our implementation of the UE receiver. chronization [45], channel estimation [44], or interfer-
After disassembling the RBs according to the UE ence awareness [46].
resource allocation, MIMO Orthogonal Frequency Divi-
sion Multiplexing (OFDM) detection is carried out. The B. Complexity
simulator currently supports Zero-Forcing (ZF), Linear Link level simulators are in practice a direct standard-
Minimum Mean Squared Error (LMMSE), and soft compliant implementation of the Physical (PHY) layer
sphere decoding as detection algorithms. The detected procedures, including segmentation, channel coding,
soft bits are decoded to obtain the data bits and several MIMO, transmit signal generation, pilot patterns, and
figures of merit, such as coded/uncoded BER, BLER, synchronization sequences. Therefore, implementation
and throughput. complexity and simulation time are in general high. To
Currently, four different types of channel estimators are obtain a simulator with readable and maintainable code,
supported within the simulator: (i) Least Squares (LS), (ii) a high-level language (MATLAB) has been chosen. This
Minimum Mean Squared Error (MMSE), (iii) Approxi- choice enabled us to develop the simulator in a fraction
mate LMMSE [44], and (iv) genie-driven (near) perfect of the time required for an implementation in other lan-
channel knowledge based on all transmitted symbols. guages such as C. Furthermore, MATLAB ensures
LTE requires UE feedback to adapt the transmission cross-platform compatibility. While MATLAB is cer-
to the current channel conditions. The LTE standard tainly slower than C, by means of code optimization
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 5 of 14
http://asp.eurasipjournals.com/content/2011/1/29

(vectorization) and parallelization by the MATLAB Par- investigated, the computational complexity of the simu-
allel/Distributed Computing Toolbox, simulation run- lation can considerably be reduced by only evaluating
time can be greatly reduced. Severely difficult-to- the user of interest. In order to enable a functional sche-
vectorize and often-called functions are implemented in duler, it is sufficient to compute just the feedback para-
C and linked to the MATLAB code by means of MEX meters for all other users. To start a simple single-cell
functions. Such functions include the channel coding/ multi-user simulation, run the file LTE_sim_batch_sin-
decoding [47], Cyclic Redundancy Check (CRC) compu- gle_cell_multi_user.m.
tation [48], and soft sphere decoding. 3) Multi-cell multi-user
Furthermore, it is possible to adjust the scale of the This simulation is by far the computationally most
simulation to the specific needs. This is achieved by demanding scenario and covers the links between multi-
introducing three different simulation types with largely ple eNodeBs and UEs. This set-up allows for the realis-
different computational complexity (Figure 4): tic investigation of interference-aware receiver
1) Single-downlink techniques [52], interference management (including
This simulation type only covers the link between one cooperative transmissions [53] and interference align-
eNodeB and one UE. Such a set-up allows for the inves- ment [54,55]), and network-based algorithms such as
tigation of channel tracking, channel estimation [44], joint resource allocation and scheduling. Furthermore,
synchronization [11,49], MIMO gains, AMC and feed- despite the vast computational efforts needed, such
back optimization [13], receiver structures [14] (neglect- simulations are crucial to verify system level simulations.
ing interference and impact of the scheduling, a To start a simple multi-cell multi-user simulation, run
modeling of channel encoding and decoding [15,50], the file LTE_sim _batch_multi_cell_multi_user.m.
and physical layer modeling [51], which can be used for The simulation time, which depends mainly on the
system level abstraction of the physical layer. To start a desired precision and statistical accuracy of the simula-
simple single-downlink simulation, run the file LTE_- tion results, the selected bandwidth, the transmission
sim_batch_single_downlink.m. mode, and the chosen modulation order, is for most
2) Single-cell multi-user users a crucial factor. It should be noted that by a smart
This simulation covers the links between one eNodeB choice of the simulation settings, the simulation time can
and multiple UEs. This set-up additionally allows for the be decreased (e.g., when investigating channel estimation
investigation of receiver structures that take into performance, the smallest bandwidth can be sufficient).
account the influence of scheduling, multi-user MIMO
resource allocation, and multi-user gains. Furthermore, 3. The Vienna LTE system level simulator
this set-up allows researchers to investigate practically In this section, we describe the overall structure of the
achievable multi-user rate regions. In the current imple- Vienna LTE System Level Simulator, currently devel-
mentation, the simulator fully evaluates the receivers of oped (January 2011) version 1.3r427. We furthermore
all users. However, if receiver structures are being show how the PHY layer procedures have been
abstracted in a low complexity manner.

A. Structure of the simulator

multi-cell multi-user In system level simulations, the performance of a whole
network is analyzed. In LTE, such a network consists of
a multitude of eNodeBs that cover a specific area in
which many mobile terminals are located and/or moving
around. While simulations of individual physical layer
links allow for the investigation of MIMO gains, AMC
feedback, modeling of the channel code, and retransmis-
X2
sions [13,44,45,50,56], it is not possible to reflect the
effects of cell planning, scheduling, or interference in a
single-downlink
large scale with dozens of eNodeBs and hundreds of
users. Simply performing physical layer simulations of
single-cell multi-user the radio links between all terminals and base-stations is
unfeasible for system level investigations because of the
Figure 4 Three possible scenarios in the Vienna LTE link level vast amount of computational power required. Thus,
simulator allow us to adjust the scale of the simulation the physical layer has to be abstracted by simplified
complexity: single-downlink, single-cell multi-user, and multi- models capturing its essential dynamics with high accu-
cell multi-user.
racy at low complexity.
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 6 of 14
http://asp.eurasipjournals.com/content/2011/1/29

Based on the standard approach in the literature

[51,57], our simulator consists of two parts: (i) a link
eNodeBs
measurement model, and (ii) a link performance model.
scheduler
The link measurement model reflects the link quality, resource allocation
channel adaptation
given by the UE measurement reports, and is required
attached UEs DL channels
to carry out link adaptation and resource allocation. The
chosen link quality measure is evaluated per subcarrier.
Based on the Signal to Interference and Noise Ratio downlink channels
(SINR), the UE computes the feedback (PMI, RI, and macroscopic pathloss
UE-specific shadow fading feedback
CQI), which is employed for link adaptation at the eNo- signaling
small scale fading UL-Channel
deB as described in Section 2-A. The scheduling algo-
signaling

interference
rithm assigns resources to users to optimize the

signal
performance of the system (e.g., in terms of throughput) UEs
based on this feedback [21]. Based on the link measure- subcarrier SINR
calculation & averaging noise
ment model, the link performance model predicts the
BLER of the link, based on the receiver SINR and the SINR-to-CQI mapping

transmission parameters (e.g., modulation and coding). SINR-to-BLER mapping BLER, throughput
Figure 5 illustrates the interaction between the two
Figure 6 Schematic class diagram showing the implementation
models and the several physical layer parameters.
of the link-to-system model in the LTE System Level Simulator
Implementation-wise, the simulator follows the struc- and depicting the relationship between the several elements/
ture shown in Figure 6. Each network element is repre- classes in the simulator. Implementation-wise, both the link-
sented by a suitable class object, whose interactions are measurement and the link-performance model reside in the UE
described below. class.
In order to generate the network topology, transmis-
sion sites are generated, to which three eNodeBs are the cell layout (defined by the eNodeB positions, large-
appended, i.e., sectors, each containing a scheduler (see scale (macroscopic, macro-scale) pathloss, shadow fad-
Figure 6). In the simulator, traffic modeling assumes full ing [58]), and the time-variant small-scale (microscopic,
buffers in the downlink. A scheduler assigns PHY micro-scale) fading [59].
resources, precoding matrices, and a suitable MCS to The CQI feedback report is calculated based on the
each UE attached to an eNodeB. The actual assignment subcarrier SINRs and the target transport BLER. The
depends on the scheduling algorithm and the received CQI reports are generated by an SINR-to-CQI mapping
UE feedback. [35] and made available to the eNodeB implementation
At the UE side, the received subcarrier post-equaliza- via a feedback channel with adjustable delay. At the
tion symbol SINR is calculated in the link measurement transmitter, the appropriate MCS is selected by the CQI
model. The SINR is determined by the signal, interfer- to achieve the targeted BLER during the transmission.
ence, and noise power levels, which are dependent on Especially in high mobility scenarios, the feedback delay
caused by computation and signaling timings can lead to
a performance degradation if the channel state changes
network layout significantly during the delay. In the link performance
base-station deployment
antenna gain pattern
model, an AWGN-equivalent SINR (gAWGN) is
interference macro-scale fading
tilt/azimuth
structure antenna gain obtained via Mutual Information Effective Signal to
shadow fading
Interference and Noise Ratio Mapping (MIESM) [60-62].
mobility management
link-measurement micro-scale fading In a second step, gAWGN is mapped to BLER via AWGN
model
link performance curves [34,35]. The BLER value acts as
precoding
traffic model
a probability for computing ACK/NACKs, which are
resource scheduling combined with the Transport Block (TB) size to compute
strategy
link-performance link adaptation strategy the link throughput. The simulation output consists of
model
power allocation strategy traces, containing link throughput and error ratios for
each user, as well as cell aggregates, from which statistical
throughput error rates error distribution
distributions of throughputs and errors can be extracted.
Figure 5 Schematic block diagram of the LTE system level
simulator. Link quality is evaluated by means of the link- B. Complexity
measurement model, while the link-performance model maps it to
One desirable functionality of a system level simulator is
BLER and outputs link throughput and error distribution.
the ability to precalculate as many of the simulation
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 7 of 14
http://asp.eurasipjournals.com/content/2011/1/29

parameters as possible. This not only reduces the com- when directly working with MIMO channel matrices on
putational load while carrying out a simulation, but also system level [16,35,51].
offers repeatability by loading an already partly precalcu-
lated scenario. 4. Validation of the simulators
The precalculations involved in the LTE system level The validation of the simulators was performed in two
simulator are the generation of (i) eNodeB-dependent steps. First, in Section 4-A we compared the link level
large-scale pathloss maps, (ii) site-dependent shadow throughput with the minimum performance require-
fading maps, and (iii) time-dependent small-scale fading ments stated by 3GPP in the technical specification TS
traces for each eNodeB-UE pair. 36.101 [65]. Second, in Section 4-B, we cross-validated
1) Pathloss and fading maps the link and the system level simulators by comparing
The large-scale pathloss and the shadow fading are their results against each other. Other means of valida-
modeled as position-dependent maps. The large-scale tion are being discussed in Section 4-C.
pathloss is calculated according to well-known models
[58,63] and combined with the antenna gain pattern of A. 3GPP minimum performance requirements
the corresponding eNodeB. Space-correlated shadow The technical specification TS 36.101 [65] defines mini-
fading is obtained from a log-normal random distribu- mum performance requirements for a UE that utilizes a
tion using a low-complexity variant of the Cholesky dual-antenna receiver. These requirements have to be
decomposition [64]. Inter-site map correlation for sha- met by real devices and therefore have to be surpassed
dow fading is similarly obtained. Figure 7 shows exemp- by our simulator, in which not every conceivable influ-
lary large-scale pathloss and shadow fading maps. ential factor is incorporated.b Such factors may include
2) Time-dependent fading trace frequency and timing synchronization as well as other
While the large-scale pathloss and the shadow fading non-ideal effects, such as quantization or non-ideality of
are modeled as position-dependent trace, the small-scale the manufactured physical components (e.g., I/Q imbal-
fading is modeled as a time-dependent trace. The calcu- ances, phase noise, and power amplifier nonlinearities).
lation of this latter trace is based on the transmitter pre- In particular, TS 36.101 specifies reference measure-
coding, the small-scale fading MIMO channel matrix, ment channels for the Physical Downlink Shared Chan-
and the receive filter. Currently, the receiver modeling is nel (PDSCH) (comprising bandwidth, AMC scheme,
based on a linear ZF receiver. The small-scale fading overhead,...) and propagation conditions (power delay
trace consists of the signal power and the interference profiles, Doppler frequencies, and antenna correlation).
power after the receive filter. The break-down into these The considered simulation scenarios are completely spe-
two parts significantly reduces the computational effort cified by referring to sections and test numbers in TS
since it avoids many complex multiplications required 36.101. For example, in TS 36.101 Section 8.2.1.1.1, the
tests for a single transmit antenna N T = 1 and dual
receive antenna NR = 2 scenario are defined. By refer-
ring to test number one in Section 8.2.1.1.1 of TS
eNodeB

36.101, the AMC mode is defined as Quadrature Phase

site-dependent: same
eNodeB-dependent: one for all eNodeBs in site Shift Keying (QPSK) with a target coding rate of 1/3,
independent pathloss map
per eNodeB od
eB
eN
od Extended Vehicular A (EVehA) channel model with a
eB
eN
Doppler frequency of 5 Hz, and low antenna correlation.
For our simulations presented in this article, we selected
pathloss [dB] shadow fading [dB] four test scenarios with a bandwidth of 10 MHz but dif-
1000 140 40
800
ferent transmit modes (single antenna port transmission,
130 30
600
20 OLSM, and TxD), different AMC schemes, and different
400 120
channel models. Hybrid Automatic Repeat reQuest
y pos [m]

200 10
110
0
100
0 (HARQ) is supported by at most three retransmissions.
−200 −10
−400 90 −20
The most important parameters of the test scenarios are
−600
−800 80 −30 listed in Table 1. The first scenario (8.2.1.1.1/1) refers to
−1000 70
−40 the test scenario described above. The OLSM scenario
−1000 −500 0 500 1000 −1000 −500 0 500 1000
x pos [m] x pos [m] (8.2.1.3.2/1) utilizes a rank two transmission, that is,
Figure 7 eNodeB- and site-dependance of the large-scale transmission of two spatial streams.
pathloss and shadow fading. Left: Large-scale pathloss and Simulation results for the considered scenarios are
antenna gain map [dB] corresponding to the lower-leftmost shown in Figure 8. The dashed horizontal lines corre-
eNodeB. Right: space-correlated shadow fading corresponding to spond to 70% of the maximum throughput values for
the site [dB].
which TS 36.101 defines a channel signal-to-noise ratio
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 8 of 14
http://asp.eurasipjournals.com/content/2011/1/29

Table 1 Test scenarios of 3GPP TS 36.101 Table 2 Test scenarios for the cross-comparison of the
8.2.1.1.1/1 8.2.1.1.1/8 8.2.1.2.1/1 8.2.1.3.2/1 link and system level simulators (SU CASE)
TX mode Single ant. Single ant. TxD OLSM SISO TxD OLSM CLSM
Channel EVehA ETU EVehA EVehA Channel TU TU TU TU
Doppler freq. 5 Hz 300 Hz 5 Hz 70 Hz Bandwidth 1.4 MHz 1.4 MHz 1.4 MHz 1.4 MHz
Modulation QPSK 16QAM 16QAM 16QAM Antenna conf. 1×1 2×2 2×2 4×2
Code rate 1/3 1/2 1/2 1/2 CQI feedback ✓ ✓ ✓ ✓
NT × NR 1×2 1×2 2×2 4×2 RI feedback ✕ ✕ ✓ ✓
Antenna corr. Low High Medium Low PMI feedback ✕ ✕ ✕ ✓
Channel SNR req. -1 dB 9.4 dB 6.8 dB 14.3 dB Simulation time LL 3 200 s 9 500 s 19 500 s 14 200 s
Simulation time SL 800 s 1 000 s 1 100 s 1 200 s
Speed-up 4 9.5 17.7 11.8
(SNR) requirement (shown as crosses in Figure 8). For TU, typical urban channel model [75].
all the considered test scenarios, the link level simulator
outperforms the minimum requirements by approxi- feedback schemes originally presented in [13]. In
mately 2-3 dB. The small vertical bars within the mar- order to create an equivalent simulation scenario on
kers in 8 are the 99% confidence intervals of the link and system level, we do not employ shadow fad-
simulated mean throughput. Since the confidence inter- ing. While on link level the SNR is usually directly
vals are much smaller than the distances between the specified, on system level the SNR is a function of
individual throughput curves, we know that a repeated the user location in the cell. Without shadow fading,
simulation with different seeds of the random number the user SNR on system level becomes a function of
generators will lead to similar results and conclusions. the distance between base-station and the user. This
Figure 8 can be reproduced by calling the script Repro- can be utilized to indirectly select appropriate SNR
ducibility_RAN_sims.m included in the link level values in the system level simulator. The results of
simulator. the link and system level comparisons are shown in
Figure 9. For all the considered simulation scenarios,
B. Link and system level cross-comparison we obtain an excellent match between the results of
Next, we cross-compare the performance of the link and the two simulators, confirming the validity of our
system level simulators. We consider a single user sin- Link Error Prediction (LEP) model [50] on system
gle-cell scenario with different antenna configurations level. Figure 9 can be reproduced by running the
and transmit modes, as summarized in Table 2. script Reproducibility _LLvsSL_batch.m provided in
Depending on the channel conditions, we adapt the the system level simulator package. Further compari-
AMC scheme, the transmission rank, and the precod- sons between link and system level simulator results
ing matrices. For this purpose, we utilize the UE are shown in Section 5-B.

25
8.2.1.3.2/1
10
link level
mean throughput [Mbit/s]

20
system level
mean throughput [Mbit/s]

8
15 4x2 CLSM 2x2 OLSM
8.2.1.1.1/8
6
1x1 SISO
10 8.2.1.2.1/1
2x2 TxD
4
5 8.2.1.1.1/1
2
0
-15 -10 -5 0 5 10 15 20 0
channel SNR [dB] -10 0 10 20 30 40
channel SNR [dB]
Figure 8 Throughput simulations of the test scenarios in 3GPP
TS 36.101 and comparison to the minimum performance Figure 9 Cross-comparison of throughput results obtained
requirements (marked with crosses). The small vertical bars with the link level and the system level simulators. The small
within the circular markers indicate the 99% confidence intervals. vertical bars within the circular markers indicate the 99% confidence
Reproducible by running Reproducibility_RAN_sims.m. intervals. Reproducible by running Reproducibility_LLvsSL_batch.m.
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 9 of 14
http://asp.eurasipjournals.com/content/2011/1/29

In Table 2 we compare the simulation times of the we present a link level throughput simulation in which
link level simulator to those of the system level simula- we compare the throughput of the different MIMO
tor. The simulations were conducted on a single core of schemes to theoretic bounds. Based on this simulation
a 2.66 GHz Quad Core CPU. The table also states the setup, researchers can investigate algorithms such as
simulation speed-up, defined as the ratio of the simula- channel estimation, detection, or synchronization. Sec-
tion times required with the link level and the system ond, we compare the performance of different state-of-
level simulator, respectively. The speed-up of the system the-art schedulers in a single-cell multi-user environ-
level simulator for a Single-Input Single-Output (SISO) ment. These schedulers serve as reference for research-
system equals four. This speed-up is rather small ers investigating advanced scheduling techniques.
because equalization, demodulation, and decoding (tasks
that are abstracted on system level) have low complexity A. Link level throughput
in a SISO system. With increasing system complexity Before presenting the link level throughput results of the
also the speed-up increases. We expected the largest different LTE MIMO schemes, we introduce theoretic
speed-up in the CLSM scenario, because it utilizes the bounds for the throughput. We identify three bounds,
largest antenna configuration. However, we measured namely, the mutual information, the channel capacity,
the largest speed-up of almost 18 in the OLSM simula- and the so-called achievable mutual information.
tion scenario. The reason is, that the precoder changes Depending on the type of channel state information
from one subcarrier to the next, while in the CLSM sce- available at the transmitter (only receive SNR, full, or
nario, we assumed wideband feedback meaning that the quantized), an ideal transmission system is expected to
same precoder is employed on all the subcarriers [13]. attain one of these bounds.
The link level simulator supports the parallel comput- 1) Mutual information
ing capabilities of MATLAB. With these features, it is The mutual information is the theoretic bound for the
possible to run several MATLAB instances in parallel data throughput if only the receive SNR but no further
on the multiple cores of a modern CPU. The simulation channel state information is available at the transmitter
time of the link level simulator then decreases linearly side [70]:
with the number of CPU cores, while the system level
simulator is currently not capable of parallel computing. 
Ntot  
1
I= Bsub log2 det INR + 2 Hk HH (1)
σn k
C. Further validation means k=1
For a basic validation of the correctness of the results where Bsub denotes the bandwidth occupied by a sin-
produced by the simulator, we checked the uncoded gle data subcarrier, Hk the NR×NT (= number of receive
BER and throughput performance over frequency flat antennas × number of transmit antennas) dimensional
Rayleigh fading and AWGN channels, as the theoretical MIMO channel matrix of the k-th subcarrier, σn2 the
performance of these channels is known [66]. Further- energy of noise and interference at the receiver, Ntot the
more, we cross-checked our results with those produced total number of usable subcarriers, and INR an identity
by the other industry simulators, by comparing with matrix of size equal to the number of receive antennas
corresponding publications of the 3GPP RAN WG1, e. NR. In Equation (1), we normalized the transmit power
g., [28,29]. Still, an open issue is to prove a correct func- to one and the channel matrix according to
tionality of each part of the simulator. Evaluation of the
E{||Hk ||22 } = 1. Therefore, Equation (1) does not show a
simulators has also been made possible for the whole
dependence on the transmit power and the number NT
research community, allowing everybody to modify the
of transmit antennas.
code to meet individual requirements and to check the
The bandwidth Bsub of a subcarrier is calculated as
code for correctness [67-69], as the simulator’s change-
log reflects. The first versions of the simulators have N
been released in May 2009 (link level simulator) and in Bsub = , (2)
Tsub − Tcp
March 2010 (system level simulator), respectively. To
facilitate the exchange of bugs and/or results often where N s is the number of OFDM symbols in one
referred to as “crowdsourcing,” a forumc is also pro- subframe (usually equal to 14 when the normal cyclic
vided. While the authors acknowledge this is not a per- prefix length is selected), Tsub the subframe duration (1
fect form of validation, neither is any other. ms), and Tcp the time required for the transmission of
all cyclic prefixes within one subframe. Note that we are
5. Exemplary results calculating the mutual information for all usable subcar-
In this section, we show two exemplary simulation riers of the OFDM system, thereby taking into account
results obtained with the Vienna LTE simulators. First, the loss in spectral efficiency caused by the guard band
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 10 of 14
http://asp.eurasipjournals.com/content/2011/1/29

carriers. If different transmission systems that apply dif- information and has to apply the optimum precoder.
ferent modulation formats are to be compared, however, Furthermore, the receiver needs to apply the optimum
a fair comparison then requires calculating the mutual receive filter to separate the parallel SISO subchannels.
information over the entire system bandwidth instead of 3) Achievable mutual information
calculating it only over the usable bandwidth. Both mutual information and channel capacity do not
Current communication systems employ adaptive consider system design losses caused, for example, by
modulation and coding schemes to optimize the data the transmission of cyclic prefix or reference symbols,
throughput. For a specific receive SNR, assuming an or the quantization of the transmitter precoding. In
optimum receiver, the modulation and coding scheme order to obtain a tighter bound for the link level
that maximizes the data throughput can be selected. throughput, we therefore consider these effects in the
Thus, if the transmitter knows the receive SNR, a definition of the so-called achievable mutual informa-
throughput equal to the mutual information should be tion. In the case of open-loop transmission, in which
achieved. space-time coding is employed at the transmitter, we
2) Channel capacity obtain for the achievable mutual information:
For calculating the channel capacity of a frequency  
selective MIMO channel [66], consider the singular (OL)

Ntot
1 1 H
Ia = FBsub log2 det INR NL + 2 H̃k H̃k , (5)
value decomposition of the channel matrix Hk scaled by NL σn
k=1
the standard deviation sn of the additive white Gaussian
noise impairment: with NL denoting the number of spatial transmission
layers. The N R N L ×NT dimensional matrix H̃k is the
1 
Hk = Uk VH ; with effective channel matrix including the space-time coding
σn k
[71]. The factor F accounts for the inherent system
k
   (3)
losses due to the transmission of the cyclic prefix and
= diag λk,m m = 1 . . . min(NR , NT )
the reference symbols. In detail, the factor F is calcu-
k
lated as
The optimum, capacity-achieving, frequency-dependent
precoding at the transmitter is given by the unitary matrix Tsub − Tcp Nsc · Ns /2 − Nref
F= · ,
Vk. If this precoding matrix is applied at the transmitter Tsub Nsc · Ns /2 (6)

 

and also the optimum receive filter UH k is employed, then CP loss reference symbols loss
the MIMO channel is separated into min(NR, NT) (with
NR denoting the number of receive antennas and NT the where N ref is the number of reference symbols per
number of transmit antennas) independent SISO channels, resource block, and Nsc = 12 is the number of subcar-
 riers in each RB. In LTE, the number of reference sym-
each with a gain of λk,m ,, m = 1...min(NR, NT), k = 1...
bols depends on the number of transmit antennas.
Ntot. The channel capacity is obtained by optimally distri-
Therefore, the efficiency factor F decreases with increas-
buting the available transmit power over these parallel
ing number of transmit antennas (see Table 3).
SISO subchannels. The optimum power distribution Pk, m
In the case of closed-loop transmission, a channel-
is the solution of the optimization problem:
adapted precoding matrix W is chosen from a set W
1  
min(NR ,NT ) Ntot (defined in the standard) and applied to the transmit
C = max log2 (1 + Pk,m λk,m ) signal. We calculate the achievable mutual information
Pk,m Ntot
m=1 k=1 for closed-loop transmission as
(4)
 
min(NR ,NT ) Ntot

Ntot 
subject to Pk,m = Pt . (CL)
Ia = max FBsub log2 det INR + 1 H H
. (7)
W∈W σn2 Hk WW Hk
m=1 k=1 k=1

where the second equation is a transmit power con- In Figure 10, the throughput of a 2×2 LTE system
straint that ensures an average transmit power equal to with 5 MHz bandwidth, perfect channel knowledge, and
the number of data subcarriers: P t = N tot . Note that

owing to the definition of λk,m , in Equation (3), the
Table 3 pilot symbols and efficiency factor F in LTE
power distribution Pk, m and thus Pt remain dimension-
Transmit antennas Reference symbols Efficiency factor F
less. We calculate the power coefficients maximizing NT Nref (%)
Equation (4) by the water-filling algorithm described in
1 4 88.88
[66]. In order to achieve a throughput equal to the chan-
2 8 84.44
nel capacity, the transmitter needs full channel state
4 12 80
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 11 of 14
http://asp.eurasipjournals.com/content/2011/1/29

In particular, we consider in the Vienna LTE System

OLSM Level Simulator one sector of a single-cell SISO system
40
with 20 randomly positioned users. The user positions
yield the large-scale path loss and shadow fading coeffi-
achievable
30 mutual cients of all the users, and as a consequence, the average
information 41%
receive SNRs, which are distributed in a range of 2.7 dB
rate [Mbit/s]

TxD
to 36 dB. These average receive SNRs of the 20 users
20 channel capacity %
4 dB 15 are set in the Vienna LTE Link Level Simulator to
1.7 dB 5.2 dB
mutual throughput ensure the same propagation environment as on system
information
10 level. Further simulation parameters of both simulators
44%
are summarized in Table 4.
0 The simulation results are averaged over 2,500 small-
-10 0 10 20 30 scale fading and noise realizations. In order to guarantee
channel SNR [dB]
exactly the same channel realizations for all scheduler
Figure 10 Throughput of a 2×2 system with 5-MHz bandwidth simulations on system level, the user positions, as well
compared to the channel capacity, the mutual information,
and the achievable mutual information. The small vertical bars
as the small- and large-scale fading realizations are
within the circular markers indicate the 99% confidence intervals. loaded from pre-generated files. On link level, the seeds
Reproducible by running Physical_Layer_batch.m. of the random number generators for fading and noise
generation are set at the beginning of each simulation.
A performance comparison of different scheduling
a Soft Sphere Decoder (SSD) receiver is shown and strategies is shown in Figures 11 and 12 in terms of
compared to the previously presented theoretic bounds. total sector throughput and fairness (Jain’s fairness
The difference between channel capacity and mutual index [73]). The figures show that the results produced
information is only small, and, therefore, even knowl- by the link and system level simulators are very similar
edge of the full channel state information at the trans- for both throughput and fairness. The largest difference
mitter does not considerably increase the potential between the results of the two simulators is less than
performance. In contrast, the difference between the 2%, while the 99% confidence intervals (too small to be
mutual information and the achievable mutual informa- identified in the figures) of the simulated throughput are
tion is much larger, resulting in a loss of 56% at an SNR much smaller. Thus, we conclude that the system level
of 15 dB. Most (41%) of this loss is due to the restric- simulator is properly calibrated; that is, the approxima-
tions implied by the standard, as indicated by the tion and modeling of the link level does not result in
achievable mutual information curves in Figure 10. At a large errors on system level.
rate of 16 Mbit/s, the difference between achievable The considered schedulers pursue different goals for
mutual information and simulated throughput is resource allocation. The best CQI scheduler tries to
approximately 4 dB. These findings are similar to the maximize total throughput and completely ignores fair-
results we obtained when analyzing the performance of ness by just assigning resources to the users with the
WiMAX and High-Speed Downlink Packet Access
(HSDPA) in [72].
Table 4 Link and system level parameters for the
Figure 10 furthermore shows that, for SNRs lower scheduling simulations
than 14 dB, the TxD mode outperforms OLSM. Only at
Parameter Value
larger SNRs, above 20 dB, where the throughput of the
System bandwidth 5 MHz
TxD mode saturates, OLSM benefits from the second
Number of subcarriers 300
spatial stream and outperforms TxD.
Number of resource blocks 50
Figure 10 can be reproduced by executing the script
Number of users 20
Physical_Layer_batch.m provided in the Vienna LTE
Channel model 3GPP TU [76]
Link Level Simulator package.
Channel realizations 2 500
Antenna configuration 1 transmit, 1 receive (1 × 1)
B. LTE scheduling
Receiver Zero forcing (ZF)
In this section, the performance of various multiuser
Schedulers Best CQI (BCQI)
LTE scheduling techniques is compared by means of
Maxmin
link level and system level simulations. By appropriately
Proportional fair
selecting the simulation parameters in the link level, as
Resource fair
well as the system level, we are able to show that the
Round robin
results obtained by the two simulators are equivalent.
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 12 of 14
http://asp.eurasipjournals.com/content/2011/1/29

proportional fair strategy outperforms resource fair in

system level link level terms of throughput as well as fairness thereby resulting
25
in a good tradeoff between throughput and fairness.
mean throughput [Mbit/s]

Further details about the implemented schedulers, as

20
well as more simulation results, can be found in [21].
The presented simulation results can be reproduced
15
by calling the script Reproducibility_Schedulers_batch.m
10
that can be found in the directory “paper scripts” of the
link level and the system level simulator, respectively.
5
More examples of the Vienna LTE simulators also in
the context of LTE-Advanced are presented in [74].
0
best CQI prop. fair res. fair maxmin round robin
scheduler 6. Conclusions
Figure 11 Comparison of system throughput obtained with In this paper, we presented the Vienna LTE Simulators,
different scheduling strategies with link and system level consisting of a link level and a system level simulator.
simulations. Reproducible by running Both simulators are available under a non-commercial
Reproducibility_Schedulers_batch.m.
open source academic-use license and thereby enable
researchers to implement and test algorithms in the
best channel conditions. This is reflected in the simula- context of LTE. The open source availability of the
tion results in Figures 11 and 12, showing the highest simulators facilitates researchers to reproduce published
system throughput and the lowest fairness for the best results in the context of LTE, and thus supports the
CQI scheduler. In contrast, the maxmin-scheduler comparison of novel algorithms with previous state-of-
assigns the resources in a way that equal throughput for the-art. So far (July 2011), the simulators have been
all users is guaranteed, thereby maximizing Jain’s fair- downloaded more than 18,000 times from all over the
ness index [73]. Round robin scheduling does not con- world.
sider the user equipment feedback and cyclically assigns
the same amount of resources to each user. Thus, ignor- Endnotes
a
ing the user equipment feedback results in the worst Note that the scheduler in a multi-user system will
throughput performance of all schedulers consider here. change the statistics of the individual user’s channel,
The proportional fair scheduler emphasizes multiuser thus influencing the receiver performance. bAfter all, the
diversity by scheduling the user who has the best cur- purpose of a simulation model is to abstract and thus
rent channel realization relative to its own average. The simplify complex situations.c http://www.nt.tuwien.ac.at/
resource fair scheduling strategy guarantees an equal forum and http://www.nt.tuwien.ac.at/forum.
amount of resources for all users while trying to maxi-
mize the total throughput. In the simulations, the Abbreviations
3GPP: third generation partnership project; AMC: adaptive modulation and
coding; ARQ: automatic repeat request; AWGN: additive white Gaussian
noise; BER: bit error ratio; BLER: block error ratio; CDF: cumulative density
system level link level function; CDMA: code-division multiple access; CoMP: cooperative multi-
1 point; CPICH: common pilot channel; CQI: channel quality indicator; CRC:
cyclic redundancy check; CSI: channel state information; DoF: degrees of
freedom; D-TxAA: double transmit antenna array; ECR: effective code rate;
0.8 EESM: exponential effective SINR mapping; FDD: frequency division duplex;
Jain’s fairness index

FFT: fast fourier transform; GSM: global system for mobile communications;
0.6
HSDPA: high-speed downlink packet access; HSPA: high-speed packet access;
HS-DPCCH: high-speed dedicated physical control channel; HSDSCH: high-
speed downlink shared channel; HS-PDSCH: high-speed physical downlink
0.4 shared channel; HS-SCCH: high-speed shared control channel; HSUPA: high-
speed uplink packet access; IMS: IP multimedia subsystem; ICI: inter carrier
interference; ISI: inter symbol interference; ITU: international
0.2
telecommunication union; LEP: link error prediction; LTE-A: LTE advanced;
LMMSE: linear minimum mean squared error; MAC-hs: medium access
0 control for HSDPA; MCS: modulation and coding scheme; MIESM: mutual
best CQI prop. fair res. fair maxmin round robin information effective SINR mapping; MIMO: multiple-input multiple-output;
scheduler MU-MIMO: multi-User MIMO; MVU: minimum variance unbiased; NACK: non-
Figure 12 Comparison of fairness obtained with different acknowledged; PCI: precoding control information; PMI: precoding matrix
scheduling strategies with link and system level simulations. indicator; PedA: pedestrian A; PedB: pedestrian B; RAN: radio access network;
Reproducible by running Reproducibility_Schedulers_batch.m. RB: resource block; RI: rank indicator; ROI: region of interest; RLC: radio link
control; RRC: radio resource control; UE: user equipment; OFDMA:
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 13 of 14
http://asp.eurasipjournals.com/content/2011/1/29

orthogonal frequency division multiple access; OFDM: orthogonal frequency 16. M Wrulich, M Rupp, Computationally efficient MIMO HSDPA system-level
division multiplexing; PDP: power delay profile; SAE: system architecture modeling. EURASIP Journal on Wireless Communications and Networking,
evolution; SCM: spatial channel model; SISO: single-input single-output; SM: 2009 (2009). Article ID 382501
spatial multiplexing; SINR: signal to interference and noise ratio; SNR: signal- 17. TM Cover, Comments on broadcast channels. IEEE Trans. Inf. Theory. 44(6),
to-noise ratio; SQP: sequential quadratic programming; ST-MMSE: space-time 2524–2530 (Oct. 1998). doi:10.1109/18.720547
MMSE; SU single user; TB: transport block; TBS: transport block size; TTI: 18. H Weingarten, Y Steinberg, S Shamai, The capacity region of the gaussian
transmission time interval; TxAA: transmit antenna array; UMTS: universal multiple-input multiple-output broadcast channel. IEEE Trans. Inf. Theory.
mobile telecommunications system; UTRA: UMTS terrestrial radio access; 52(9), 3936–3964 (Sept. 2006)
WCDMA: wideband CDMA; ZF: zero-forcing. 19. U Niesen, P Gupta, D Shah, The balanced unicast and multicast capacity
regions of large wireless networks. IEEE Trans. Inf. Theory. 56(5), 2249–2271
Acknowledgements (May 2010)
This work has been funded by the Christian Doppler Laboratory for Wireless 20. D Skoutas, D Komnakos, D Vouyioukas, A Rouskas, Enhanced dedicated
Technologies for Sustainable Mobility, KATHREIN-Werke KG, and A1 Telekom channel scheduling optimization in WCDMA, in Proc. 14th European Wireless
Austria AG. The financial support by the Federal Ministry of Economy, Family Conference (EW) (June 2008)
and Youth and the National Foundation for Research, Technology and 21. S Schwarz, C Mehlführer, M Rupp, Low complexity approximate maximum
Development is gratefully acknowledged. The authors would like to thank throughput scheduling for LTE, in Conference Record of the 44th Asilomar
Christoph F. Mecklenbräuker for his valuable comments and fruitful Conference on Signals, Systems and Computers, Pacific Grove, CA, USA
discussions. (Nov. 2010)
22. K Peppas, T Al-Gizawi, F Lazarakis, D Axiotis, A Moussa, A Alexiou, System
Competing interests level evaluation of reconfigurable MIMO techniques enhancements for
The authors declare that they have no competing interests HSDPA, in Proc. IEEE Global Telecommunications Conference (GLOBECOM). 5,
2869–2873 (2004)
Received: 5 November 2010 Accepted: 25 July 2011 23. P Gkonis, D Kaklamani, G Tsoulos, Capacity of WCDMA multicellular networks
Published: 25 July 2011 under different radio resource management strategies, in Proc. 3rd International
Symposium on Wireless Pervasive Computing (ISWPC), pp. 60–64 (2008)
References 24. M Castañeda, M Ivrlac, J Nossek, I Viering, A Klein, On downlink intercell
1. M Barni, F Perez-Gonzalez, Pushing science into signal processing. IEEE interference in a cellular system, in Proc. IEEE 18th International
Signal Process. Mag. 22(4), 120–119 (July 2005) Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC),
2. P Vandewalle, J Kovačević, M Vetterli, Reproducible research in signal pp. 1–5 (2007)
processing. IEEE Signal Process. Mag. 26(3), 37–47 (May 2009) 25. T Nihtila, V Haikola, HSDPA MIMO system performance in macro cell
3. M Dohler, R Heath Jr, A Lozano, CB Papadias, RA Valenzuela, Is the PHY network, in Proc. IEEE Sarnoff Symposium (2008)
layer dead?. IEEE Communications Magazine, 4, 159–165 (Apr. 2011) 26. KI Pedersen, TF Lootsma, M Stottrup, F Frederiksen, T Kolding, PE
4. JB Buckheit, DL Donoho, Wavelab and reproducible research. Dept. of Mogensen, Network performance of mixed traffic on high speed downlink
Statistics, Stanford Univ., Tech. Rep. 474, Tech. Rep, http://www-stat.stanford. packet access and dedicated channels in WCDMA, in Proc. IEEE 60th
edu/~donoho/Reports/1995/wavelab.pdf (1995) Vehicular Technology Conference (VTC). 6, 4496–4500 (2004)
5. P Vandewalle, S Süsstrunk, M Vetterli, A frequency domain approach to 27. SA Jafar, MJ Fakhereddin, Degrees of freedom for the MIMO interference
registration of aliased images with application to super-resolution. EURASIP channel. IEEE Trans. Inf. Theory. 53(7), 2637–2642 (July 2007)
Journal on Applied Signal Processing, 2006 (2006). Article ID 71459 28. Nortel, Performance evaluation of CL MIMO under different UE speed. 3rd
6. C Mehlführer, S Caban, M Rupp, Experimental evaluation of adaptive Generation Partnership Project (3GPP), Tech. Rep. R1-080384 (Jan. 2008)
modulation and coding in MIMO WiMAX with limited feedback. EURASIP 29. Nokia & Nokia-Siemens-Networks, LTE performance benchmarking. 3rd
Journal on Advances in Signal Processing, Special Issue on MIMO Systems with Generation Partnership Project (3GPP), Tech. Rep. R1- 071960 (Apr. 2007)
Limited Feedback, 2008 (2008). Article ID 837102 30. steepest ascent, 3GPP LTE toolbox and blockset, (2009)
7. M Šimko, D Wu, C Mehlführer, J Eilert, D Liu, Implementation Aspects of 31. mimoOn, mi!mobile, (2009)
Channel Estimation for 3GPP LTE Terminals, in Proc. European Wireless 32. Aricent, LTE layer 1 - LTE baseband/PHY library, (2009)
(EW 2011), Vienna, Austria (Apr. 2011) 33. JJ Sánchez, G Gómez, D Morales-Jiménez, JT Entrambasaguas, Performance
8. R Dallinger, H Ruotsalainen, R Wichman, M Rupp, Adaptive pre-distortion evaluation of OFDMA wireless systems using WM-SIM platform, in Proc.
techniques based on orthogonal polynomials, in Conference Record of the ACM 4th International Workshop on Mobility Management and Wireless
44th Asilomar Conference on Signals, Systems and Computers, Pacific Grove Access (MobiWac), pp. 131–134 (2006)
(CA), USA (Nov. 2010) 34. C Mehlführer, M Wrulich, JC Ikuno, D Bosanska, M Rupp, Simulating the
9. Technical Specification Group RAN, E-UTRA; LTE physical layer - general long term evolution physical layer, in Proc. 17th European Signal Processing
description. 3GPP, Tech. Rep. TS 36.201 Version 8.3.0 (March 2009) Conference (EUSIPCO 2009) Glasgow, Scotland, UK, pp. 1471–1478
10. C Williams, S McLaughlin, MA Beach, Exploiting multiple antennas for (Aug. 2009)
synchronization. IEEE Trans. Veh. Technol. 58(2), 773–787 (Feb. 2009) 35. JC Ikuno, M Wrulich, M Rupp, System level simulation of LTE networks, in
11. Q Wang, S Caban, C Mehlführer, M Rupp, Measurement based throughput Proc. 2010 IEEE 71st Vehicular Technology Conference (VTC2010-Spring),”
evaluation of residual frequency offset compensation in WiMAX, in Proc. Taipei, Taiwan (May 2010)
51st International Symposium ELMAR-2009, Zadar, Croatia pp. 233–236 36. Technical Specification Group RAN, E-UTRA; physical channels and
(Sept. 2009) modulation. 3GPP, Tech. Rep. TS 36.211 Version 8.7.0 (May 2009)
12. N Kolehmainen, J Puttonen, P Kela, T Ristaniemi, T Henttonen, M Moisio, 37. Technical Specification Group RAN, E-UTRA; multiplexing and channel
Channel quality indication reporting schemes for UTRAN long term coding. 3GPP, Tech. Rep. TS 36.212 (March 2009)
evolution downlink, in Proc. IEEE Vehicular Technology Conference Spring 38. Technical Specification Group RAN, E-UTRA; physical layer procedures. 3GPP,
(VTC), pp. 2522–2526 (May 2008) Tech. Rep. TS 36.213 (March 2009)
13. S Schwarz, C Mehlfuhrer, M Rupp, Calculation of the spatial preprocessing 39. X Wang, G Giannakis, A Marques, A unified approach to QoS-guaranteed
and link adaption feedback for 3GPP UMTS/LTE, in Proc. IEEE Wireless scheduling for channel-adaptive wireless networks, in Proceedings of the
Advanced,” London, UK (June 2010) IEEE. 95(12), 2410–2431 (Dec. 2007)
14. L Boher, R Legouable, R Rabineau, Performance analysis of iterative receiver 40. CL Raymond Kwan, J Zhang, Multiuser scheduling on the downlink of an
in 3GPP/LTE DL MIMO OFDMA system, in Proc. IEEE 10th International LTE cellular system. Research Letters in Communications. 2008 (2008). Article
Symposium on Spread Spectrum Techniques and Applications (ISSSTA), pp. ID 323048
103–108 (Aug. 2008) 41. S Schwarz, C Mehlführer, M Rupp, Throughput maximizing multiuser
15. M Wrulich, M Rupp, Performance and modeling of LTE H-ARQ, in Proc. scheduling with adjustable fairness, in Proc. IEEE International Conference on
International ITG Workshop on Smart Antennas (WSA 2009), Berlin, Germany Communications (ICC 2011), Kyoto, Japan (May 2011)
(Feb. 2009)
Mehlführer et al. EURASIP Journal on Advances in Signal Processing 2011, 2011:29 Page 14 of 14
http://asp.eurasipjournals.com/content/2011/1/29

42. Members of ITU, Recommendation ITU-R M.1225: Guidelines for evaluation 65. Technical Specification Group RAN, Evolved universal terrestrial radio access
of radio transmission technologies for IMT-2000. International (E-UTRA); user equipment (UE) radio transmission and reception. 3GPP,
Telecommunication Union (ITU), Tech. Rep (1997) Tech. Rep. TS 36.101 Version 8.5.1 (Mar. 2009)
43. L Hentilä, P Kyösti, M Kaske, M Narandzic, M Alatossava, MATLAB 66. A Paulraj, R Nabar, D Gore, Introduction to Space-Time Wireless
implementation of the WINNER phase ii channel model ver1.1, http://www. Communications, 1st edn. (Cambridge Univ. Press, 2003)
ist-winner.org/phase_2_model.html (Dec. 2007) 67. Y Feng, A Krewski, WL Schroeder, Simulation based comparison of metrics
44. M Šimko, C Mehlführer, M Wrulich, M Rupp, Doubly dispersive channel and measurement methodologies for OTA test of MIMO terminals, in Proc.
estimation with scalable complexity, in Proc. International ITG Workshop on of the Fourth European Conference on Antennas and Propagation (EuCAP
Smart Antennas (WSA 2010), Bremen, Germany, pp. 251–256 (Feb. 2010) 2010) (Apr. 2010)
45. Q Wang, C Mehlführer, M Rupp, Carrier frequency synchronization in the 68. VH Muntean, M Otesteanu, G-M Muntean, QoS parameters mapping for the
downlink of 3GPP LTE, in Proc. 21st Annual IEEE International Symposium on e-learning traffic mix in LTE networks, in Proc. International Joint Conference
Personal, Indoor and Mobile Radio Communications (PIMRC 2010), Istanbul, on Computational Cybernetics and Technical Informatics (ICCC-CONTI 2010),
Turkey (Sept. 2010) pp. 299–304 (May. 2010)
46. M Šimko, C Mehlführer, T Zemen, M Rupp, Inter carrier interference 69. D Wu, J Eilert, R Asghar, D Liu, VLSI implementation of a fixed-complexity
estimation in MIMO OFDM systems with arbitrary pilot structure, in Proc. soft-output MIMO detector for high-speed wireless. EURASIP Journal on
73rd IEEE Vehicular Technology Conference (VTC2011-Spring), Budapest, Wireless Communications and Networking. 2010 (2010). Article ID 893184
Hungary (May 2011) 70. C Mehlführer, S Caban, M Rupp, Measurement-based performance
47. I Solutions, Iterative Solutions Coded Modulation Library (ISCML), http:// evaluation of MIMO HSDPA. IEEE Transactions on Vehicular Technology,
www.iterativesolutions.com/ accepted for publication
48. T Pircher, pycrc CRC calculator and C source code generator, http://www. 71. B Badic, M Rupp, H Weinrichter, Adaptive channel-matched extended
tty1.net/pycrc/ alamouti space-time code exploiting partial feedback. ETRI Journal. 26(5),
49. Q Wang, C Mehlführer, M Rupp, SNR optimized residual frequency offset 443–451 (2004). doi:10.4218/etrij.04.0703.0006
compensation for WiMAX with throughput evaluation, in Proc. 17th 72. M Rupp, C Mehlführer, S Caban, On achieving the Shannon bound in
European Signal Processing Conference (EUSIPCO 2009), Glasgow, Scotland, cellular systems, in Proc. 20th International Conference Radioelektronika 2010,”
UK (Aug. 2009) Brno, Czech Republik (Apr. 2010)
50. JC Ikuno, C Mehlführer, M Rupp, A novel link error prediction model for 73. R Jain, D Chiu, W Hawe, A Quantitative Measure of Fairness and
OFDM systems with HARQ, in Proc. IEEE International Conference on Discrimination for Resource Allocation in Shared Computer Systems. DEC,
Communications (ICC 2011), Kyoto, Japan (May 2011) Tech. Rep. TR-301 (Sept. 1984)
51. M Wrulich, S Eder, I Viering, M Rupp, Efficient link-to-system level model for 74. S Caban, C Mehlführer, M Rupp, M Wrulich, Evaluation of HSDPA and LTE,
MIMO HSDPA, in Proc. of the 4th IEEE Broadband Wireless Access Workshop,” From Testbed Measurements to System Level Performance, Wiley, (2011)
New Orleans, LA, USA (Dec. 2008) 75. 3GPP, Technical Specification Group Radio Access Networks; Deployment
52. C Mehlführer, M Wrulich, M Rupp, Intra-cell interference aware equalization aspects (Release 8), http://www.3gpp.org/ftp/Specs/html-info/25943.htm
for TxAA HSDPA, in Proc. 3rd IEEE International Symposium on Wireless (Dec. 2008)
Pervasive Computing (ISWPC 2008), Santorini, Greece, pp. 406–409 76. Technical Specification Group Radio Access Network, Deployment aspects
(May 2008) (Release 8). 3GPP, Tech. Rep. TS 25.943 Version 8.0 (Dec. 2008)
53. A Ibing, V Jungnickel, Joint transmission and detection in hexagonal grid
for 3GPP LTE, in Proc. International Conference on Information Networking doi:10.1186/1687-6180-2011-29
2008 (ICOIN 2008) (Jan. 2008) Cite this article as: Mehlführer et al.: The Vienna LTE simulators -
54. V Cadambe, S Jafar, Interference alignment and degrees of freedom of the Enabling reproducibility in wireless communications research. EURASIP
K-user interference channel. IEEE Trans. Inf. Theory. 54(8), 3425–3441 Journal on Advances in Signal Processing 2011 2011:29.
(Apr. 2008)
55. R Tresch, M Guillaud, Cellular interference alignment with imperfect channel
knowledge, in IEEE International Conference on Communications Workshops
2009 (ICC Workshops 2009) (June 2009)
56. P Wu, N Jindal, Performance of hybrid-ARQ in block-fading channels: A
fixed outage probability analysis. IEEE Transactions on Communications.
58(4), 1129–1141 (Apr. 2010)
57. Y Blankenship, P Sartori, B Classon, V Desai, K Baum, Link error prediction
methods for multicarrier systems, in Proc. IEEE 60th Vehicular Technology
Conference (VTC2004-Fall). 6, 4175–4179 (Sept. 2004)
58. Technical Specification Group RAN, E-UTRA; LTE RF system scenarios. 3GPP,
Tech. Rep. TS 36.942 (2008)
59. T Zemen, C Mecklenbräuker, Time-variant channel estimation using discrete
prolate spheroidal sequences, in IEEE Trans. Signal Process. 53(9), 3597–3607
(Sept. 2005)
60. J Kim, A Ashikhmin, A van Wijngaarden, E Soljanin, N Gopalakrishnan, On
efficient link error prediction based on convex metrics, in Proc. IEEE 60th
Vehicular Technology Conference (VTC2004-Fall). 6, 4190–4194 (Sept. 2004)
61. S Tsai, A Soong, Effective-SNR mapping for modeling frame error rates in
multiple-state channels. 3GPP2, Tech. Rep. 3GPP2-C30-20030429-010
Submit your manuscript to a
(Apr. 2003) journal and benefit from:
62. L Wan, S Tsai, M Almgren, A fading-insensitive performance metric for a
unified link quality model, in Proc. IEEE Wireless Communications and 7 Convenient online submission
Networking Conference (WCNC2006). 4, 2110–2114 (Apr. 2006) 7 Rigorous peer review
63. Technical Specification Group RAN, Physical layer aspects for E-UTRA. 3GPP, 7 Immediate publication on acceptance
Tech. Rep. TS 25.814 (2006)
7 Open access: articles freely available online
64. H Claussen, Efficient modelling of channel maps with correlated shadow
fading in mobile radio systems, in Proc. IEEE 16th International Symposium 7 High visibility within the field
on Personal, Indoor and Mobile Radio Communications (PIMRC2005). 1, 7 Retaining the copyright to your article
512–516 (Sept. 2005)
Submit your next manuscript at 7 springeropen.com