A Software Defined Radio Platform with Raspberry Pi and Simulink

A Software Deﬁned Radio Platform

with Raspberry Pi and Simulink

Gianni Pasolini, Flavio Zabini

Wilab, University of Bologna

Bologna (Italy)

Email: [email protected]

ﬂ[email protected]

Alessandro Bazzi

IEIIT-CNR

Bologna (Italy)

Email: [email protected].it

Stefano Olivieri

The MatWorks Inc.

Turin (Italy)

Email: stefano.oli[email protected]

Abstract—The Raspberry Pi is a low cost single-board com-

puter that gained recently the attention of hobbyists and practi-

tioners, especially for ﬁle server and media server applications.

Thanks to the free support package released by MathWorks, it

is fully supported by Simulink, hence it can be programmed

according to the model-based design paradigm. This means that

also users with no programming skills can design and implement

their own applications on Raspberry Pi boards.

In this paper we show how Raspberry Pi’s boards can be used,

jointly with Simulink, to easily implement digital communication

systems. This allows the realization of practical experiences,

intended for use within telecommunication courses, that provide

a viable low-cost way to introduce students to the software-deﬁned

radio paradigm.

I. INTRODUCTION

The last decade witnessed an impressive development of

programmable microchips speciﬁcally designed for signal pro-

cessing tasks, such as ﬁeld programmable gate arrays (FPGAs)

and digital signal processors (DSPs). Such devices made the

software deﬁned radio (SDR) concept a solid reality, so that

nowadays the baseband section of a communication system

can be implemented by the software running on such hardware.

Modern telecommunication engineers need, therefore, a

broad expertise in the SDR ﬁeld, encompassing programming

languages, digital signal processing techniques and hardware-

related issues, which should be primarily acquired during their

classes at the University.

Unfortunately, although telecommunication laboratory fa-

cilities are usually available within Electronic and Telecom-

munication Faculties, practical SDR experiences are seldom

proposed within courses, for two main reasons:

• the cost of the hardware (DSP and FPGA development

kits), that can hardly be replicated in many working

stations and, above all, left in the hands of inexperienced

users,

• the complexity of such experiences, which require a

ﬁrst teaching phase focused on communication-systems

design and digital signal processing aspects, followed by

a second phase devoted to the basic of DSP and FPGA

programming, and a third practical phase dedicated to the

hardware implementation and the measurements.

In many cases, therefore, the time reserved to practical SDR

experiences (if any) is a very limited fraction of the duration

of digital signal processing courses, that are more focused on

theoretical aspects.

Both the above issues may be overcome, however, by the

recent introduction in the mass market of general purpose

programmable devices with two fundamental characteristics:

• a very limited cost (few dozens of dollars),

• the possibility to automatically generate the software

starting from the functional model of the system to

be implemented, according to the model based design

(MBD) paradigm.

The Raspberry Pi board [1] is, perhaps, the most relevant

example of such devices: it is a popular low cost single-board

computer developed by the Raspberry Pi Foundation with the

intention of promoting the teaching of basic computer science.

It is largely used as a server in home networks or as a media

center and it is also adopted in many academic courses [2].

What is particularly relevant for the educational purpose

addressed here, is that the Raspberry Pi is fully supported by

MathWorks Inc. through the Raspberry Pi Support Package

that establishes a tight connection between the Raspberry Pi

and Simulink [3]. This “marriage” allows the realization of

a powerful, yet simple, SDR educational platform. Simulink

allows, in fact, the graphical modeling and the simulation of

the system to be implemented, and then translates the model

into software, which is downloaded on the device, thanks to the

automatic code generation carried out by the Support Package.

This is making possible the realization of the SDR practical

experiences that we are collecting in [4], which don’t require

any programming skills and can be implemented at a very lim-

ited cost. Students facing the experimental activities, equipped

with the Raspberry Pi as well as one PC hosting Simulink and

the basic instrumentation of a telecommunication laboratory

(signal generator, oscilloscope, ...), will be able to experience

the full life-cycle of a project, from the design and the

simulation to the hardware implementation and the subsequent

measurements. This is what we call Simulink deﬁned radio.

In the following a description of the SDR educational

platform is provided, along with an example of experimental

activity already developed.

Hardware Support Packages are speciﬁc add-ons that enable the use of

MathWorks product with third-party hardware.

2016 24th European Signal Processing Conference (EUSIPCO)

Fig. 1. Raspberry Pi 2 Model B.

(a)

External

sound card.

(b) Micro SD

memory card.

LAN adapter.

(d) USB-micro

USB cable.

(e) LAN ca-

ble.

(f) Stereo

3.5mm-RCA

jack cable.

(g) RCA-

BNC adapter.

Fig. 2. Additional hardware.

II. THE RASPBERRY PI

The Raspberry Pi 2 Model B, which is the 2nd generation

Raspberry Pi model shown in Fig. 1, is a credit-card sized

single-board computer equipped with a quad-core Broadcom

BCM2836 ARM v7 processor running at 900MHz. Despite its

low cost, in the order of 40$, it features 1GB RAM, a 40 pin

GPIO connector, four USB ports, a 4-pole stereo output and

composite video port, a HDMI port, a CSI connector, a DSI

connector, a micro-SD card slot and an Ethernet socket.

It gained the attention of hobbyists and practitioners espe-

cially for ﬁle server and media server applications. Nowadays,

there are plenty of existing projects, easily available on the

Internet, which work on 1st generation or 2nd generation Rasp-

berry Pis. Expert users can also develop their own applications

using Python or other programming languages, taking beneﬁt

from the availability of such a versatile and cheap device to

realize customized solutions to their needs.

In this paper we show an unconventional usage of Rasp-

berry Pi boards: deprived of input/output peripherals such

as monitor, keyboard, and mouse, and added of the cables

and connectors summarized in Fig. 2 (and better discussed in

Section IV-A), they can be used as DSPs for the realization

of communication systems and signal processing subsystems.

To achieve this goal, both an analog input and an analog

output are required. Since the Raspberry Pi does not have

Fig. 3. Comparison between the external DAC output and the Raspberry Pi 2

Model B headphones output. 10 kHz sine wave.

natively an analogue output, we used an external USB sound

card with the 33051D chipset, like the one shown in Fig. 2(a),

that supplies a microphone input and one more headphones

output.

This cheap device (about 20$), connected to the Rasp-

berry Pi’s USB port, actually plays a double role: analog to

digital converter (ADC) for the input signals and digital to

analog converter (DAC) for the output signals. As the device is

conceived for audio signals, its sampling frequency is limited

to 48000 samples per second, with a resolution of 16 bits

per sample. It follows that the band of generated and received

signals must be within the interval [0, 24 kHz]. This should not

be deemed a problem, however, because the signal bandwidth

is not really relevant for teaching purposes. In the case of

digital transmissions, for instance, the transmitter architecture

does not change at all, and it simply entails that the bit rate

must be kept low.

Observe, ﬁnally, that although the Raspberry Pi is equipped

with an integrated audio output (headphones output), it is

surely preferable to use the analog output provided by the

external sound card. The integrated DAC is, in fact, a low cost

component and the poor quality of its output can be observed

for example in Fig.3, where the yellow curve represents a 10

kHz sine generated by a Raspberry Pi and measured at the

integrated output, whereas the red curve represents the same

signal measured at the sound card output. As clearly evident,

the signal generated by the built-in DAC is quite distorted.

III. SIMULINK AND THE MODEL BASED DESIGN

PARADIGM

Simulink, developed by MathWorks, is a graphical exten-

sion to MATLAB for modeling and simulation of linear and

nonlinear dynamic systems.

In Simulink, systems are drawn on screen by means of

simple drag-and-drop operations of elementary blocks, that are

interconnected each other to realize the ﬁnal block diagram.

The elementary blocks are collected in a comprehensive library

of toolboxes, that includes also virtual input and output devices

such as function generators and oscilloscopes.

2016 24th European Signal Processing Conference (EUSIPCO)

399

Fig. 4. Picture of the workstation.

Simulink is, in particular, the basic tool for MBD, which

is an efﬁcient and cost-effective way to develop complex

systems in few steps that go from the realization of their

functional model (the block diagram) to the ﬁnal hardware

implementation, through intermediate steps that include sim-

ulation, automatic code generation, and continuous test and

veriﬁcation [5].

Thanks to the free Support Package provided by Mathworks,

Raspberry Pi boards are fully supported by Simulink, which

makes MBD a powerful methodology to easily get to the

implementation of actual systems on such devices. At the end

of the modeling and simulation phases, in fact, Simulink is

capable of translating the designed model into an executable,

which is ﬁnally downloaded on the board.

From the perspective of the signal processing teaching

methodology, this possibility is with no doubt a revolution.

It permits, in fact, the hardware implementation of complex

systems by simply drawing the corresponding block diagram

with Simulink, thus allowing:

• an immediate visual correspondence with the functional

block diagrams explained during theoretical lessons,

• the possibility to observe and measure the actual signals

generated/received/processed by the implemented system,

• to get rid of the need to teach programming languages,

focusing the attention on signal processing aspects only,

• to reduce the cost of realization of an SDR laboratory.

IV. SIMULINK DEFINED RADIO WITH RASPBERRY PI

Based on Raspberry Pi 2 Model B (hereafter simply referred

as Rasperry Pi) and Simulink, several SDR experiences were

realized in strict cooperation between Mathworks and the

University of Bologna to teach the modeling and subsequent

hardware implementation of basic telecommunication systems

and signal processing algorithms. The complete platform and

a single example experience are hereafter described. More

details and all examples are accessible at [4].

A. Workstation Setup

The software, the laboratory equipment, and the additional

hardware needed to setup the workstation based on the Rasp-

berry Pi are detailed hereafter (see Figs. 4 and 5).

Fig. 5. Block scheme of the workstation.

1) Software: The SDR experimental activities proposed in

this paper require, ﬁrst of all, a PC equipped with MATLAB

and Simulink. In particular, the experiences mentioned below

have been realized with MATLAB R2015a equipped with the

following libraries:

• Signal Processing Toolbox;

• DSP System Toolbox;

• Communications System Toolbox.

2) Laboratory equipment: The experimental activities re-

alized by means of the SDR platform here discussed result

in the realization of systems able, in general, to generate and

process signals.

The system modeling stage, carried out using Simulink,

and the consequent Raspberry Pi implementation, are thus

followed by a signal measurement phase, aimed at verifying

the correct functioning of the system designed and to conﬁrm,

through experimental observations, the theoretical concepts

concerning the system implemented.

The equipment required for the measurement campaign is

typically available in every didactic lab for electronics and

telecommunications, with particular reference to:

• Signal generator. It will be mostly used as a sine wave

generator, in order to provide the carrier needed by some

of the transmitters implemented, or the input signal for

digital ﬁltering systems.

• Oscilloscope. In the majority of cases, the observation of

signals will focus on their behaviour in the time domain.

This task is performed through an oscilloscope, that will

be employed in most of the experimental activities.

• Spectrum analyser. Allows to investigate the power

distribution of a signal along the frequency axis.

It is worth noting that the signals that will be gener-

ated/processed by our systems are within the [0 24 kHz]

band, owing to the characteristics of the external sound card

introduced in Section II. For the generation or analysis of such

signals there is no need for sophisticated instruments; the basic

instruments typically available in a didactic lab, generally solid

but not highly performing, are suitable for the experimental

activities described below.

In the cases when even basic oscilloscopes and spectrum

analyzers are deemed too costly, an interesting alternative

2016 24th European Signal Processing Conference (EUSIPCO)

400

Fig. 6. Simulink model of the OFDM transmitter.

is the use of the Digilent Analog Discovery device, which

is a low cost multi-purpose instrument conceived for low

frequency applications. When connected to the PC through

the USB port, this device is able to generate and acquire

signals. With a moderate price, in the order of 270$, this

tool can operate as signal generator, oscilloscope, spectrum

analyzer, network analyzer, logic state analyzer, digital signal

generator, and power supply. The user interface of each single

instrument, displayed on the PC’s monitor, shows the same

knobs, sliders and buttons of the full hardware instrument,

allowing the user to perform the measurement activity as

if he were in a lab. Of course, the bandwidth that can be

handled by the Digilent Analog Discovery, in the order of some

MHz, cannot be compared with that of more sophisticated and

expensive instruments, however it is more than adequate for

the didactic experiences carried out with our SDR platform.

In this case, therefore, the signal generator, the oscilloscope

and the spectrum analyzer can be conveniently replaced by

this multifunction tool.

3) Additional hardware: In addition to the Raspberry Pi,

the PC hosting Simulink, and the laboratory equipments, the

hardware depicted in Fig. 2 is needed:

• Nr. 1 External sound card (Fig. 2(a)). As explained in

Section II, it is used to provide the Raspberry Pi with an

analog input, otherwise absent, and with a higher quality

analog output.

• Nr. 1 Micro SD memory card (Fig. 2(b)). It contains

the Raspberry Pi operating system.

• Nr. 1 USB to LAN adapter (Fig.2(c)). If the PC’s

Ethernet port is not available, this adapter allows to

connect the Raspberry Pi to the PC using the USB port.

• Nr. 1 USB-micro USB cable (Fig. 2(d)). It is used to

connect the Raspberry Pi to the PC’s USB port, with the

only aim to provide the power supply.

• Nr. 1 Network cable (Fig. 2(e)). It is used to connect

the Raspberry Pi to the PC.

• Nr. 2 3.5 mm-RCA jack cables (Fig. 2(f)). They are

used to connect the sound card, equipped with 3.5 mm

female jacks (Fig. 2(a)), to the instruments. RCA-BNC

adapter (Fig. 2(g)) are also needed.

• Nr. 2 RCA-BNC adapter (Fig. 2(g)). They are used to

connect the 3.5 mm-RCA jack cable to the instruments,

usually equipped with BNC connectors (Fig.2(f)).

The cost of all the additional hardware remains below 60$.

The basic workbench setup is shown in Fig. 5: one PC

hosting Simulink connected, through an Ethernet link, to the

Raspberry Pi, whose output is fed to an oscilloscope and/or

a spectrum analyzer. Some experiences also require a signal

generator, whose task is to provide the Raspberry Pi with an

external signal (such as the carrier).

B. Developed experiences

As already stated, several SDR experiences are being devel-

oped in the framework of a cooperation between Mathworks

and the University of Bologna, using the above described plat-

form. In particular, the following systems have been already

implemented:

• Signal generator;

• 2-PAM and 4-PAM baseband transmitters;

• 2-ASK and 4-ASK transmitters;

• FSK transmitter;

• Q-PSK transmitter;

• OFDM transmitter (64 and 256 carriers);

• Digital ﬁltering;

• Adaptive noise canceler.

The corresponding Simulink models and the related documen-

tation, conceived as material to be provided to students, can

be freely downloaded at [4]. For the sake of conciseness, only

the example case of an OFDM transmitter is here presented.

V. EXAMPLE: RASPBERRY PI AS AN OFDM TRANSMITTER

To provide an example case, the Simulink model that has

been developed to implement an OFDM transmitter is shown

in Fig. 6.

To simplify the description, the following seven

macro-blocks have been highlighted in the ﬁgure:

1) Symbol generation: The Symbol generation macro-

block generates real symbols a

∈ {−1, 1}, starting from

the bits produced by the Bernoulli Binary Generator block.

Such symbols will be used to modulate the subcarriers of the

OFDM signal; as the alphabet used has just two symbols, the

modulation adopted for the subcarriers is 2-ASK. The bit rate

is set to 1800 bits/s.

In order to keep the model as simple as possible we did not implement

the cyclic-preﬁx.

2016 24th European Signal Processing Conference (EUSIPCO)

401

2) OFDM modulator: Starting from the symbols a

its input, the OFDM modulator macro-block generates the

baseband signal corresponding to an OFDM modulation with

N = 64 subcarriers, N

= 48 of which are actually used to

transmit modulation symbols (useful subcarriers) whereas the

remaining N

= 16 have null amplitude (15 virtual subcarriers

+ 1 DC subcarrier in correspondence of the frequency zero

More speciﬁcally, the N = 64 subcarriers are used as follows:

• 8 virtual subcarriers (from 1 to 8) with null amplitude;

• 24 data subcarriers (from 9 to 32) with 2-ASK modula-

tion;

• 1 DC subcarrier (number 33) with null amplitude (corre-

sponding to the zero frequency);

• 24 data subcarriers (from 34 to 57) with 2-ASK modu-

lation;

• 7 virtual subcarriers (from 58 to 64) with null amplitude.

With the introduction of the virtual and DC subcarriers, the

output sampling rate of the OFDM macro-block is 2400 sam-

ples/s. Data are organized in frames to simplify the real time

management by the hardware.

3) Upsampling: In order to modulate the baseband OFDM

signal, the sampling frequency must be increased. It is conve-

nient, on this regard, to increase it by a factor of 20, taking it

to the 48000 samples/s required by the audio card’s DAC. The

upsampling operation is performed by the sequence of blocks

included in the Upsampling macro-block, that performs an

upsampling by a factor of 2 at ﬁrst and then by a factor of

10. The two stage procedure is preferable, compared with a

single stage upsampling by a factor of 20, because it reduces

the overall computational burden [6], [7].

4) Baseband to IF: The Baseband to IF macro-block mod-

ulates the signal, translating it from baseband to intermediate

frequency. Such operation is performed through a quadrature

modulator with internally generated carriers at a frequency

= 15 kHz. The Baseband to IF macroblock is a classic

quadrature modulator, with two separate paths for the real

(in-phase) and the imaginary (quadrature) components of the

signal. Each component is upconverted by a mixer (represented

by the Product block) driven by a cosine or a sine carrier,

generated by the Cosine Wave and Sine Wave blocks. Both

modulated signals are then summed up and taken out.

5) Automatic Gain Control: The Automatic Gain Control

macro-block adapts the signal’s dynamic range to the require-

ments of the output port. As a consequence of this operation,

the highest value in each frame is equal to 2

− 1, consistent

with the highest value required by the output port (2

− 1).

6) Raspberry Pi output: The Raspberry Pi output macro-

block represents the system output port. It corresponds, there-

fore, to the sound card’s DAC.

Once the Simulink model is realized and checked through

Simulink simulations, it is possible to carry out the Deploy

to Hardware, which is triggered by Simulink itself. The

In order to facilitate the receiver in the research of the band center, the

subcarrier corresponding to the zero frequency (in the baseband) is usually

assigned a null amplitude. The acronym DC means Direct Current.

Fig. 7. OFDM signal spectrum.

implemented system starts as soon as the automatci download

of the corresponding software on the device is completed. Con-

necting the Raspberry Pi to the spectrum analyzer, according

to the scheme in Fig. 5, the signal spectrum appears as shown

in Fig. 7. The expected bandwidth of approximately 2 kHz

and the DC subcarrier centered at 15 kHz can be observed.

VI. CONCLUSION

Despite its popularity, the usage of Raspberry Pi boards as a

signal processing device for SDR applications is an innovative

exception and opens new possibilities to the teaching of signal

processing. Owing to its low cost, it can be massively used

in lab activities that do not require any programming skills,

thanks to the Simulink support. In the next future, even RF

transmitters and receivers could be part of the SDR platform

here presented, thank to the availability of SDR peripherals

such as the HackRF One [8] transmitter/receiver and the RTL-

SDR receiver [9].

ACKNOWLEDGMENT

The authors wish to thank Mirko Mirabella for his great

contribution to the Simulink Deﬁned Radio project.

REFERENCES

[1] Raspberry Pi website. [Online]. Available: http://www.raspberrypi.org

[2] Raspberry Pi Academy website. [Online]. Available:

http://www.raspberrypi.org/picademy/

[3] Raspberry Pi Support from Mathworks website. [Online]. Available:

http://it.mathworks.com/hardware-support/raspberry-pi-simulink.html

[4] Raspberry pi based SDR experiences. [Online]. Available:

http://www.simulinkdeﬁnedradio.com/

[5] Simulink and model based design. [Online]. Available:

http://it.mathworks.com/services/consulting/proven-solutions/model-

based-design.html

[6] R. Crochiere and L. Rabiner, “Interpolation and decimation of digital

signals; a tutorial review,” Proceedings of the IEEE, vol. 69, no. 3, pp.

300 – 331, March 1981.

[7] G. Pasolini and R. Soloperto, “Multistage decimators with minimum

group delay,” in IEEE International Conference on Communications

(ICC), May 2010, pp. 1–6.

[8] Hackrf One. [Online]. Available: https://greatscottgadgets.com/hackrf/

[9] RTL-SDR. [Online]. Available: http://www.nooelec.com/store/sdr/sdr-

receivers/nesdr-mini-rtl2832-r820t.html

2016 24th European Signal Processing Conference (EUSIPCO)

402