SDR implementation of an OFDM transmitter and

receiver
Jaime Luque Quispe∗ , Jackelyn Tume Ruiz∗ , Luis G. Meloni
Faculty of Electrical Engineering
State University of Campinas - UNICAMP
{luque, jtumer, meloni}@decom.fee.unicamp.br

Abstract—The new level of connectivity among machines and the functionality of the systems relies on the software
people promised by the forthcoming 5G mobile communication implemented in a programmable device: system on chip
system will shape diverse application areas from healthcare and (SOC) field programmable gate array (FPGA), digital signal
education to smart factories and autonomous driving. These
applications have a variety of different sets of requirements on processor (DSP) or general purpose processor (GPP). Thus,
data rates, latency, and reliability which are expected to be the reconfiguration of the system is possible, even during the
supported by an enhanced version of the OFDM modulation execution time, by reloading a firmware or modifying the
used today by the physical layer of 4G LTE systems. The software and recompiling it. Hence, SDR opens opportunities
development of a new OFDM-based air interface able to flexibly in both researching and teaching enabling the designing and
support that diversity of applications is of current interest. On
the other hand, software defined radio is an appealing concept testing of digital communication systems in a very practical
to introduce flexibility in the radios as the functionality of the way taking into account not only the above-mentioned real
communication system relies on the software implemented in channel conditions of the channel but also the front-end
a programmable device. Aiming to analyze the factibility of impairments.
implementing elements of new communication systems under an
SDR approach, this work presents the prototyping of a SDR
OFDM communication link for video transmission using the
open-source tool GNU radio and the RF hardware interface The objective of this work is to show that the SDR technol-
Universal Software Radio Peripheral. Preliminary results shows ogy is ready to be employed in the designing and prototyp-
that a SDR opens opportunities in both researching and teaching ing of digital signal processing (DSP) modem algorithms to
enabling the designing and testing of digital communication build reconfigurable communication systems. A reconfigurable
systems in a very practical way taking into account the real
conditions of the channel and front-end impairments. platform is of particular interest since building blocks can
be tailored for different applications, e.g. vehicle-to-vehicle
I. INTRODUCTION (V2V) communications can require not only short frames
Over the past years, the development of wireless com- to achieve low latency but also enlarged subcarrier spacing,
munications systems has been largely influenced by the in- narrowband internet of things (IoT) can require the use of a
creasing of the traffic demand. Emerging new applications of single carrier waveform to achieve sufficient coverage with
low latency, higher connection density and ubiquitous gigabit low power, and custom cyclic prefix (CP) would be needed to
connectivity require more efficient and flexible modem and achieve adequate service coverage in different application cell
information processing algorithms thus motivating an intense sizes. With this aim, this work presents the implementation
research and development work at industry and academia on of SDR communication link for video transmission using the
the physical layer (PHY) and medium access control (MAC) GPP of a standard PC for the digital baseband functions and
layer components of new communication systems, since they the universal software radio peripheral (USRP) for the RF
affects all the upper layers and overall system performance processing. The system uses the orthogonal frequency division
in terms of energy efficiency, spectral efficiency, achievable multiplexing (OFDM) multicarrier modulation scheme as an
throughput, quality of service (QoS), etc. On the other hand, example case since OFDM is widely used in many current
the prototyping of a communication system with ongoing communication systems such as LTE and 802.11 and probably
research needs to be flexible such that it can be able to also in 5G new radio systems. As the software layer it was
support new algorithms and test different use cases. Nowadays, used GNU radio which is a open-source set of libraries of
5G transmission and reception blocks are being formulated baseband algorithms written in C++ and python.
theoretically and analyzed by simulation, however there is a
lack of implementations that allow the proof of concepts and The remainder of this work is structured as follows: A basic
evaluation under real channel conditions, which deteriorates model of a OFDM communication system is presented in
the signal quality. section II. Section III presents the SDR implementation of
Software defined radio (SDR) is an appealing approach this model and presents some performance results. Section IV
for flexible prototyping of communication systems where presents our conclusions.
 source-coded 0
bit stream
packet
Mapper subcarrier CP
IFFT formation 10
allocator insertion
MUX
pilots
20
guard subcarriers preamble generator

Fig. 1: Block diagram of an OFDM transmitter 30

40

II. SYSTEM MODEL 50

OFDM is a multicarrier modulation scheme based on the 60

3 2 1 0 1 2 3

division of a serial data stream in a group of parallel data

streams which are transmitted on orthogonal subcarriers. By Fig. 2: Effect of pulse shaping on the reduction of the OOBE
properly configurating the OFDM waveform, when it is sent of a OFDM signal. The worst case β = 0 is included for
through a time-dispersive channel or frequency selective fad- comparison.
ing, the subcarriers can be affected only by a constant channel
or frequency flat fading. In this kind of channels, the received
signal can be distorted by intersymbol interference (ISI), can be better contained by using a raised cosine (RC) pulse
which is a crosstalk between signals arrived with different shaping filter and in this case, assuming the frequency response
delays. This problem is addressed in OFDM by inserting a of the filter to be P (f ), the PSD results in
copy of the last part of the symbol, also known as cyclic N −1 i  2
prefix (CP), at the beginning to absorb channel delay spreads. 1 X h 2  k 
P SDx (f ) = E |Sk,l | P f − . (3)
The CP helps to preserve the synchronization and maintain T TN 
k=0
the subcarrier orthogonality, and it also introduces a cyclic
In Fig. 2 is shown the reduction of the OOBE of a 5-MHz
convolution between the transmitted symbol and the impulse
LTE waveform achieved by the application of some values of
response of channel thus enabling the use of one-tap equalizers
roll-off length β of the RC filter.
to compensate the channel distortion. A CP-OFDM signal of
Additionally, in packet-based data transmissions, special
N subcarriers can be expressed in the time domain by
preamble OFDM symbols can be prepended in the front
of every frame (one frame is composed of a number of
∞ N −1 OFDM symbols). These symbols can be used for detection,
time and frequency synchronization and initial channel state
X X
x[n] = Sk,l p[n − lNT ]ej2πk(n−lNT )/N (1)
l=−∞ k=0
estimation [gnuradio]. Only for a sake of illustration, the
preamble insertion is indicated at the end of the Fig. 1. At
where N , NCP and NT are the number of samples of the this point, the OFDM signal is ready for conversion to the
payload, CP and entire OFDM symbol respectively, i.e., NT = RF domain and further transmission. These processes usually
NCP + N . involve additional digital signal processing such as digital up
A block diagram of a transmitter is shown in Fig.1, where a conversion (DUC) and digital pre-correction of possible IQ
stream of bits is converted into complex data symbols by the imbalance of the RF components.
constellation mapper and then placed on determined positions At the receiver side, two important process are needed
of a N -order inverse fast Fourier transform (FFT) by the before data demodulation, namely the synchronization, which
subcarrier allocation block. The subcarrier allocator also places detects the beginning of the OFDM symbols and estimates
the pilots for channel estimation purposes and leaves unused the frequency offset, and the channel estimation, which serves
some subcarrier positions (guard subcarriers) to contain the to compensate the distortions inserted by the channel. After
out-of-band emission (OOBE) from the active subcarriers. these processes and the discarding of the CP, the time-domain
Then, groups of N frequency-domain samples are converted to samples are converted into the frequency domain and from
the time domain by the inverse FFT and then a CP is appended them the payload data symbols are extracted and then equal-
at the beginning of each OFDM symbol. Optionally, these CP- ized using the channel response previously estimated. The
OFDM symbols can be pulse shaped in the time-domain for equalized data symbols (complex numbers) are then converted
spectral containment. For example, the power spectrum density to bits and passed to a forward error correction (FEC) block
(PSD) of 1 given by to recover the original information.
A. Synchronization
N −1 2 Synchronization is a critical stage of the reception process.
i  
k A poor synchronization accuracy leads to severe degrada-
X h
2 
P SDx = T E |Sk,l | sinc T f −
  , (2)
TN  tion of link performance thus making impossible to recover
k=0
 sync. filtered
y(t) r[n] RX pilot channel
ADC sync. DFT channel
samples estimate channel
m estimate
Timing Frequency LPF (.)-1
LO compensation
NCO
Estimation Estimation known
reference symbols
Fig. 3: Block diagram of the time and frequency OFDM
synchronization stages Fig. 4: Channel estimation and equalization processes

transmitted information. The synchronization aims to correct pilot subcarriers, they usually use low order modulations (e.g.
any timing and frequency deviations of the received signal BPSK or QPSK), for easy recovering. The main advantage
respect to the transmitted one. It also includes the sampling of this scheme is its simplicity, but unlike blind estimation
clock synchronization. Time synchronization ensures that the techniques, it loses spectrum efficiency. The estimation and
receiver samples only useful OFDM payload samples thus compensation processes using this approach are represented
avoiding ISI from adjacent OFDM symbols while frequency in Fig. 4.
synchronization removes the frequency errors (e.g. due to
III. SDR COMMUNICATION SYSTEM
Doppler effect or receiver oscillator mismatching) to guarantee
PROTOTYPING
the orthogonality of the subcarriers and thus avoiding ICI. A
block diagram of a synchronizer is shown in Fig.3. This section details about an implementation of the system
By assuming that the transmit carrier frequency is fc and presented above using USRP RF interface cards and the open-
the received signal y(t) is down-converted by a non-matched source tool GNU Radio through its graphical interface GNU
local oscillator of frequency fLO , a carrier frequency error radio companion (GRC).
(CFO) is produced whose value is fo = fc − fLO . Then, the
A. Hardware employed
sampled signal r[n] can be expressed as
There were utilized the following modules:
r[n] = y[n]ej2πfo nTs , (4)
• Two standard PC with Linux operating systems and GNU
where Ts is the sampling frequency and n is the sample radio installed. One PC acts as the transmitter and the
index. ∆m is the deviation in samples of the received signal other as the receiver.
respect to an ideal signal whose beginning is aligned with • Two USRP boards B210, which are equipped with digital-
the start of sampling. ∆m and fLO are estimated in the to-analog converters (DACs) and analog-to-digital con-
timing and frequency estimation blocks respectively. Given verters (ADCs), 2x2 MIMO antennas, reference clocks
an estimated frequency offset of fˆo , a numerically controlled and RF chip transceivers with continuous frequency cov-
oscillator (NCO) rotates the signal by exp(j2π fˆo nTs ). Then, erage from 70 MHz to 6 GHz and 56MHz of bandwidth.
the data subcarriers can be recovered by means of a FFT for They also have 3.0 USB interface for communicating
subsequent processing. In order to track online the input signal with the host PCs. This implementation used 2 GHz as
and compensate the errors as soon as possible, the synchro- the carrier frequency.
nizer needs to respond immediately [DIEGO BARRAGAN]
B. Description
which requires fast signal processing algorithms. They can
be classified in non data-aided and data-aided synchronization The GRC schematic containing the building blocks of the
algorithms whose difference is that the first scheme exploits OFDM transmitter is shown in Fig. 5 and the receiver in Fig.6.
the redundancy of the CP to perform the time and frequency In the OFDM transmitter, it can be seen that there are
offset estimation while the second one utilizes time and defined variables like sampling rate, Sync word 1, Sync word
frequency resources as preamble and pilot subcarriers. 2, occupied carriers, pilot symbols, etc, which will be used in
the transmiter and receiver OFDM. The information comes
B. Channel estimation and equalization of one TS video, indicated in the File source block, this
As mentioned above, the radio channel distorts the trans- information is sent at Stream to Tagged Stream block that
mitted signal varying its amplitude and phase, thereby the will add tags at evenly spaced intervals, the output information
response of the channel needs to be estimated to compensate passes through an error corrector CRC32 to then be divided
(equalize) the errors introduced in the OFDM frame. into Headers bits and Payload bits, which will be multiplexed.
There are different techniques to perform channel estimation Was used BPSK and QPSK modulations respectively. The
which can be classified into blind channel estimation and output information will be sent to OFDM Carrier Allocator
data-aided estimation techniques. The first scheme does not block, here is indicated the occupied carried and the Cyclic
require dedicated time slots and reference subcarriers (training Prefixed is added.
symbols or pilot subcarriers) for channel tracking, so the The information will be sent using a USRP Block, where
spectrum efficiency can be preserved. Similar to data-aided are configured the basic parameters for the transission, like
frequency estimation, channel estimation can employ reference sample rate, frequency, gain, etc. Was used a 2GHz frequency
information multiplexed with the useful payload. In case of and 2MHz sample rate.
 Fig. 5: Block diagram of the transmissor system

Fig. 6: Block diagram of the receptor system

In the OFDM receiver, we can visualize four important information receiver is saved in a output File, we obtained the
parts. First, the USRP source and the synchronizer, was used TS video sended in the transmission.
a Schmidl & Cox OFDM synchronizer to realize the frame
offset correction, immediately the samples are ducted into the
Header/Payload Demox block and a trigger signal is sent to
indicate the begining of the frame. Then, is performed the
channel estimation into the header, the FFT shifts OFDM sym-
bols to the frequency domain and the output information is sent
to the Channel Estimation block. This block carrying out the
channel estimation and the frequency offset of the preamble
samples OFDM. The firts two symbols are of syncronization
which are used to estimated the equalization coefficient, then
this symbols are discarded and the rest of symbols are sent out
of the block. When the stream of OFDM symbols is sent at Fig. 7: Physic implementation of transmission and reception
the OFDM Frame Equalizer, it realize the equalization in one system
or two dimensions for one OFDM frame, besides correct the
coarse frequency offset of the carrier. The OFDM Serializer
select the data symbols of the OFDM symbols and emits them
as a sequency of complex values, which will be demodulated. In the Fig. 7 can be see the hardware implemented to
The bits of the header then are interpreted by the Packet realize the proof of concept, booth computers have instaled the
Header Parser block, the results of the this interpretation is graphics interface of the GNU Radio and the are connected
feedback to the demuxer for lets it know the length of the to their corresponding USRP. The left computer is doing the
payload and the tagged stream. The payload demodulation information transmission, in this case the information is a
don’t need to do the estimation channel because the channel TS video, in the other hand, the right computer is doing the
state information is on the payload tagged stream. Finally, the signal reception and can be seen that a good quality video was
received as well as the OFDM spectrum.
 IV. C ONCLUSION
Increasingly, the SDR systems are introducing into radio
communications world, proporcing scalables solutions easy to
adapt to sundry kind of escenaries.
In the present work was implemented a example case
simple of undertand in orden to serve like introduction for
the construccion of more sophisticated platforms solutions
aimed at development of solutions oriented to 4G, LTE or
IoT systems.
This work provided a comprensive literature review of one
of several SDR solutions: GNU Radio platform and USRP
hardware. Based on the OFDM concepts was test out that
SDR allows desing flexible and scalables systems, capable to
adapt to sundry communications sistems and new applications.
Also, was review the most important concepts to consider
in the blocks used in the desing of the diagram blocks of
the solution such as channel model, synchronism, channel
estimation which are directly related to implementation of
OFDM system in hardware.
The realized example case, where was implemented a video
transmission and reception system using OFDM modulation
system, has been successfully, the design was did in GNU
Radio and the USRP was used like a RF interface.
R EFERENCES
[1] J. Mitola, Software Radios, Survey, Critical Evaluation and Future
Direction
[2] A. R. Bahai, B. Saltzberg, M. Ergen ,Multi-Carrier Digital Communica-
tions Theory and Applications of OFDM Second Edition
[3] https://www.ettus.com/product/details/UB210-KIT
[4] https://wiki.gnuradio.org/index.php/Main Page
[5] M. Dillinger, K. Madani, N. Alonistioti, Software Defined Radio, Systems
and Functions
[6] B. Farhang-Boroujeny, H. Moradi OFDM Inspired Waveforms for 5G ,
2016