Real Time Voice Transmission Darshana Shah

Real Time Voice transmission over Wireless medium using

GNU Radio and USRP

EEC 687 Mobile Computing

Project Report

By
Darshana Shah
Department of Electrical & Computer Engineering
Cleveland State University, OH
d.shah44@csuohio.edu
Submitted to Dr. Chansu Yu
06/07/2009

-1 -
Real Time Voice Transmission Darshana Shah

Abstract

My project is to transmit real time voice over wireless medium. For that I used two USRPs (one
for transmission and one for reception) as hardware platform and GNU Radio as software
environment for encoding and modulating voice at transmitter side and for decoding and
demodulating voice at receiver side. First section is introduction. Second section is background of
the project. Third section shows project details. It has three subsections. They are project goal,
experimental setup, past work and results. Fourth section is Discussion. Fifth section is future
work and sixth and last section is References.

1. Introduction

SDR (Software Defined Radio) implements everything in software. E.g. filter, amplifier,
encoder/decoder, modulator/demodulator etc. Free SDR tool kit is available online i.e. GNU
Radio. It has three major advantages over hardware radio. First advantage is SDR is cheap
compared to hardware because we do not need to buy encoders, modulators and all other
components. Second advantage is debugging is easy compared to hardware because if we run the
program in software it shows error message so we can easily find it and correct it and thus it
saves time. On the other hand in hardware we need to take out each component one by one and
need to test it until we find the error, so it becomes quite lengthy and time consuming process.
Third advantage is we can modify code in GNU radio software and make it according to our
requirement. It can transmit large data or voice files using different hardware E.g. USRP. I use
GNU Radio as software environment and USRP as hardware platform to transmit real time voice
from one place to other over wireless medium.

2. Background of the project

GNU Radio

It is software defined radio. It has everything in software rather than hardware. It brings software
code as close as possible to the antenna. In my project I use it to perform encoding, assembling
and channel modulation at transmitter side and decoding, disassembling and channel
demodulation at receiver side. It is available online and it is free. It uses C classes to implement
different signal processing functions and python for main application programming. Swig is used
as a glue to connect it all together. Swig is a Linux package that converts C classes to python
compatible classes.

USRP

It supports wide range of frequencies for wireless communication as one USRP mother board can
support up to four daughter boards. Thus it can cover frequency from analog range to some GHz.
Following several kinds of daughter boards are available.

BasicTX -- 2 MHz to 200 MHz Transmitter

BasicRX -- 2 MHz to 300+ MHz Receiver

-2 -
Real Time Voice Transmission Darshana Shah

LFTX -- DC-30 MHz Transmitter

LFRX -- DC-30 MHz Receiver
TVRX -- 50 MHz to 870 MHz Receiver
DBSRX -- 800 MHz to 2.4 GHz Receiver
RFX400 -- 400-500 MHz Transceiver
RFX900 -- 800-1000MHz Transceiver
RFX1200 -- 1150 MHz - 1450 MHz Transceiver
RFX1800 -- 1.5-2.1 GHz Transceiver
RFX2400 -- 2.3-2.9 GHz Transceiver, 20+mW output

I use two RFX2400 daughter boards of 2.4GHz one at transmitter side and one at receiver
side. Figure 1a shows hardware of USRP.

Figure 1a
Hardware Diagram of USRP

USRP has following main blocks.

RF front end, ADC/DAC, FPGA and Programmable USB controller.

How it works?

In this section I will explain how GNU Radio and USRP work together. It is shown in figure 1b.
We need to install GNU Radio software in computers. Sender sends the signal from GNU Radio
software installed in computer after applying required functions from program. Through USB

-3 -
Real Time Voice Transmission Darshana Shah

cable we can connect PC to USRP. The first block of USRP is FPGA which is programmable.
The function of FPGA is to up convert the signal and make it suitable to DAC. DAC converts
digital signal which is user code to analog signal and send it to RF front end. RF front end sets the
frequency of the signal suitable to antenna and transmits it to antenna.

At receiver side antenna captures signal and sends it to RF front end. RF front end makes signal
suitable to ADC by changing its frequency. ADC converts analog signal to digital and send it to
FPGA. FPGA down converts the signal. Now through USB cable we can apply this digital signal
to GNU driver which is installed in the PC. So we can receive signal from the USRP and can
apply functions of GNU Radio and observe it from PC.

3. Real Time Voice Transmission over Wireless Medium

This part includes three subsections. 3.1 introduce project goal and block diagram of the project.
3.2 show experimental setup. 3.3 covers past work and finally 3.4 shows result.

3.1 Project Goal

My project goal is to transmit real time voice through wireless medium. Basic idea is shown in
figure 2. For that I use two USRP connected to two different PCs one at transmitter side and other
at receiver side. I installed GNU Radio software in both PCs. My project goal is to speak
something in mike attached with one pc at transmitter side and it should be hear from speaker
attached to other PC at the same time. So I need encoder at transmitter side which encodes voice
and decoder at receiver side to decode voice. I use GSM-FR encoder and decoder at transmitter

-4 -
Real Time Voice Transmission Darshana Shah

side and receiver side respectively. It is already available in GNU Radio tool kit. I use
tx_voice.py and rx_voice.py from GNU Radio examples for this project. It is for real time voice
application. GSM FR was the first digital speech-coding standard which was used in GSM digital
mobile phone system. I use GMSK modulation and demodulation. In GMSK modulation signal is
first applied to Gaussian filter and then to the modulator. Its spectral efficiency is high means it
can transmit faster than QPSK and other modulation, which is necessary for real voice
transmission.

3.2 Experimental Setup

Figure 3 Project Setup

-5 -
Real Time Voice Transmission Darshana Shah

Figure 3 shows experimental setup of the project. I have connected one USRP at transmitter side
to which mike is connected and one USRP at receiver side to which speaker is connected.

3.3 Past Work

I had run different audio files and USRP files from GNU Radio examples and it works
fine.

I tried to do data communication between two USRP and it was also working. For that I
used benchmark_tx.py and benchmark_rx.py from GNU Radio examples.

I tried to transfer real time voice using tx_voice.py and rx_voice.py. But it was not
working. I always got false packets and sometimes did not receive anything.

3.4 Results

Now I am able to transmit real time voice over wireless medium. If I speak in
microphone connected to one PC I can hear it from speaker connected to other PC.

I tried to play online Internet radio on one PC and I can hear it from other PC. So I can
play any online voice application using this project.

4. Discussion

At the time of midterm project presentation I got false packets and I was not able to transmit real
time voice. So I thought there is some frequency matching problem so it does not work. Then I
tried different daughter boards but got the same problem. Then I tried to test both USRP one by
one by running some audio files from GNU Radio examples and it was working. So I got
confused what to do. I tried to use different RF channel antenna. I got many errors to make this
program run and I solved it one by one. Finally it starts working. I was able to transmit real time
voice and then I am thinking about to transmit online Internet radio. It is also little bit tricky and I
got true result means I am able to transmit online radio.

5. Future Work

We can make compact version of USRP like Ethernet card and can replace it because
USRP is compatible with GNU Radio and it is programmable so we can do lots of things
compared to Ethernet card. Obviously not for home application because USRP is costly
but at industry level it becomes very useful.

Transmission over wireless medium using router needs file sharing. That is security hole.
Once computer port becomes open hijacker can easily hijack it. If you want to access file
system without file sharing using routers we need to buy software which is very costly
still not programmable. Using USRP there is no security hole. No file sharing is needed.
So for secured data and voice communication we can use this application. We can
program USRP using FPGA according to our requirement but can not program router
accept some parameter settings.

-6 -
Real Time Voice Transmission Darshana Shah

6. References
[1] http://www.wu.ece.ufl.edu/projects/softwareRadio/#Project%20paper
[2] http://users.ecel.ufl.edu/~zhifeng/project/GNU_Radio_USRP/index.htm
[3] Software Defined Radio: with GNU Radio and USRP by Cory Clark
[4] http://www.gnuradio.org/trac
[5] Lecture notes of Dr. Yu, which is available at,
http://academic.csuohio.edu/yuc/mobile09/09_lab04_week05_usrp_ipaq.pdf
[6] http://www.ettus.com/
[7] http://www.gnu.org/software/gnuradio/doc/exploring-gnuradio.html
[8] http://en.wikipedia.org/wiki/GMSK#Gaussian_minimum-shift_keying

Appendix

I have used tx_voice.py and rx_voice.py from GNU Radio examples for this project. If you get
any problem while running this project please let me now.

-7 -