Build requirements
In order to build the gr-ofdm module you will need to install the following prerequisites
- ZeroMQ including the C++ and Python bindings (http://zeromq.org/)
- UHD Driver for USRP hardware
- GNU Radio (https://gnuradio.org/)
- ITPP (https://sourceforge.net/projects/itpp/)
- python-qwt5-qt4
- python-scipy
How to build and install
In the gr-ofdm folder do
mkdir build
cd build
cmake ../
make
Installation is optional
make install
for USRP1 devices copy the special FPGA image
usrp/images/std_1rxhb_1txhb.rbf
to the uhd images directory. This is necessary to obtain a flat frequency response.
How to run the apps
Assuming that the module has been compiled but not installed, in the gr-ofdm folder do
cd apps
./run_benchmark_gui.sh
to run the simulation or
./run_app.sh ../python/tx.py -f2.45G
And
./run_app.sh ../python/rx.py -f2.45G --disable-ctf-enhancer --scatterplot
to run the rf transmission. Then to observe everything in th graphical user interface:
./run_app.sh gui/ofdm_tx_gui.py
and
./run_app.sh gui/ofdm_rx_gui.py --tx-hostname=<hostname>
be careful with the frequency offset of you radio frontend. It might be necessary to apply a small manual correction of a few kHz.
Additional features:
Coding
./run_benchmark_gui.sh --coding --data-blocks=10
Adaptive OFDM
./run_benchmark_gui.sh --sinr-est --multipath
with USRPs and adaptive OFDM enabled
./run_usrp_tx_gui.sh --rx-hostname=<hostname or ip> -f 2.45G --tx-gain=<gain>
./run_usrp_rx_gui.sh --tx-hostname=<hostname or ip> -f 2.45G --sinr-est
Error
cmake ..error:
GNURADIO_TRELLIS_FOUND = TRUE
Checking for GNU Radio Module: UHD
-- Checking for module 'gnuradio-uhd'
-- Found gnuradio-uhd, version 3.7.13.5
*INCLUDES=/usr/local/include
*LIBS=/usr/local/lib/libgnuradio-uhd.so;/usr/local/lib/libgnuradio-runtime.so;/usr/local/lib/libgnuradio-pmt.so
-- Found GNURADIO_UHD: /usr/local/lib/libgnuradio-uhd.so;/usr/local/lib/libgnuradio-runtime.so;/usr/local/lib/libgnuradio-pmt.so
GNURADIO_UHD_FOUND = TRUE
-- Found PythonInterp: /usr/bin/python (found version "2.7.12")
-- Found NumPy: version "1.11.0" /usr/lib/python2.7/dist-packages/numpy/core/include
-- Found NumPy: version "1.11.0"
/usr/lib/python2.7/dist-packages/numpy/core/include
CMake Error at cmake/Modules/FindITPP.cmake:46 (MESSAGE):
Could not find ITPP library
Call Stack (most recent call first):
CMakeLists.txt:117 (find_package)-- Configuring incomplete, errors occurred!
See also "/home/nlj/gr-ofdm/build/CMakeFiles/CMakeOutput.log".
install itpp:
ITPP https://sourceforge.net/projects/itpp/
IT++ Configuration and Installation Instructions
================================================
Assuming that you have already downloaded the latest IT++ sources, untar and
unpack the sources, and enter the unpacked directory. Depending on the
package type you have downloaded, use the following commands:
% gzip -cd itpp-<VERSION>.tar.gz | tar xf -
% cd itpp-<VERSION>
% bzip2 -cd itpp-<VERSION>.tar.bz2 | tar xf -
% cd itpp-<VERSION>
Since version 4.3.0, the IT++ library uses cmake compilation system for
preparing Makefiles, so the compilation procedure is as follows:
% mkdir build && cd build
% cmake ..
The `cmake' command can be invoked with additional switches and options
(see cmake help for a full list of them). The most important are:
o `-DITPP_SHARED_LIB=off', allows to compile the static version of IT++ library. By default this option is
set to `on'.
o `-DBLA_VENDOR=vendor', where vendor can be `ACML', `Intel11' or `ATLAS'. This is helpful in selecting a specific
external library (ACML or MKL 11). If these libraries are installed in nonstandard locations, on Linux, set
`LD_LIBRARY_PATH' environment variable to the location where the libraries are installed. For Windows `PATH' environment
variable should be set, while Mac OS X users should use `DYLD_LIBRARY_PATH'.
o `-DGTEST_DIR=/path/to/gtest/sources' is used to specify the path to
Google unit test framework (gtest) sources. Thus the unit tests are compiled and can be
run separately to check IT++ library.
o `-DOLD_TESTS=on' is used to activate the compilation of older unit tests (no unit test framework required). By default
this option is set to `off'. These tests are no longer maintained, so it is recommended to use the tests based on Google
framework. In order to run these tests use `check_tests.py' script found in extras folder (Python is required).
o `-DHTML_DOCS=off' allows to disable the generation of HTML documentation. Default option is `on' provided that
Doxygen is found.
o `-DCMAKE_BUILD_TYPE=Release' allows to specify the build type: Release (default) or Debug. The library name in Debug
mode is itpp_debug.*, so that both Release and Debug versions could exist in the installation folder.
o `-DCMAKE_INSTALL_PREFIX=/install/path' allows to specify the installation path. This is used when installing IT++,
either from command line, with 'make install', either using the INSTALL project from Visual Studio. Note that the user
should have write permissions to the installation folder.
External Libraries
------------------
By default, cmake checks for a few external libraries, which
might be used by the IT++ library (cf. IT++ Requirements). The detection
procedure is as follows:
1. First, the presence of a BLAS library among MKL, ACML, ATLAS and NetLib's
reference BLAS is checked. If one of the above mentioned can be used,
HAVE_BLAS is defined.
2. Next, some LAPACK library is being searched, but only if BLAS is
available. Full set of LAPACK routines can be found in the MKL, ACML and
NetLib's reference LAPACK libraries. Besides, ATLAS contains a subset of
optimised LAPACK routines, which can be used with NetLib's LAPACK library
(this is described in the ATLAS documentation). If some LAPACK library
can be found, HAVE_LAPACK is defined.
3. Finally, a set of separate checks for FFT libraries is executed.
Currently three different libraries providing FFT/IFFT routines can be
used: MKL, ACML and FFTW. If at least one of them is found, HAVE_FFT id
defined. Besides, one of the following: HAVE_FFT_MKL, HAVE_FFT_ACML or
HAVE_FFTW3 is defined, respectively.
If some external libraries are installed in a non-standard location in your
system, e.g. MKL in `/opt/intel/mkl/9.1', cmake will not
detect them automatically. In such a case, you should use LD_LIBRARY_PATH
environment variable to define additional directories to be
searched for libraries and headers. For instance, to configure IT++ to link to 32-bit version
of MKL 11.0 external libraries, you should use the following commands:
% export LD_LIBRARY_PATH=/opt/intel/composer_xe_2013.2.146/mkl/lib/intel64/
% cmake .. -DBLA_VENDOR=Intel11
Compilation
-----------
Now, it is time for compiling and linking the IT++ library. To do so, please
simply run the following command:
% make
IT++ should compile without any errors or warnings. If this is not the case,
please report the problem on the IT++ Help forum at SourceForge. Please
include information about your OS, compiler version, external libraries and
their versions, etc.
Testing the Compiled Library
----------------------------
It is recommended that you check if your library has been compiled and linked
properly and works as expected. To do so, you need to specify the path to the
Google C++ Test Framework sources as:
% cmake .. -DGTEST_DIR=/path/to/gtest
% make
% gtests/itpp_gtests
As a result, you should obtain a report with test results.
If not all tests passed please report the problem on the IT++ Help forum.
Installation
------------
Finally, you should install the compiled and linked library, include files
and HTML documentation (optionally) by typing:
% make install
Depending on the CMAKE_INSTALL_PREFIX settings during configuration, you might need the
root (administrator) access to perform this step.
Eventually, you might invoke the following command
% make clean
to remove all files created during compilation process. Alternatively you could remove
build folder:
% cd ..
% rm -rf build