timina/cxlb-ephyl repository overview

⁠GR-EPHYL: GRC resources used in the EPHYL project

⁠What is it ?

• Here you can find all GNU Radio blocks and designs used in EPHYL project⁠ (Enhanced Physical Layer for Cellular IoT).
• The blocks are developed in Python (gr_modtool).
• After properly installing the module, have a look at the examples/ directory

⁠Requirements

• Ubuntu 16, Debian 8.10 or higher
• CorteXlab Toolchain 3.7⁠ already installed with at least these packages:
• UHD
• UHD Firmware
• GNU Radio v3.7
• Matlab Runtime 2014a (Default installation, no custom directories). Unless you do not need TurboFSK blocks, in this case have a look at the branch "no_turbofsk"
• Some basic knowledge of GNU Radio

⁠Installation

⁠Docker Image

Have a look at the DockerHub repository⁠. Or simply run the following:

docker pull timina/cxlb-ephyl:latest
docker run -dti --net=host timina/cxlb-ephyl

Then connect to the container via ssh:

ssh -Xp 2222 root@localhost

The main advantage of using our docker image is avoiding all installing/upgrading/downgrading steps in order to use the module. Furthermore, the docker image provides a full compatibility with the CorteXlab⁠ testbed.

⁠Local machine installation

• Be sure to install resources via the Toolchain⁠.
• Be sure to know your install directory used in the toolchain, and put it in an environment variable, in order to use later:
  • For example if you installed the toolchain using this command:

./cxlb-build-toolchain /A/B/compiled_stuff  /D/installed_stuff

• You should run this:

echo 'export INSTALL_PATH=/D/installed_stuff' >> ~/.bashrc

• If you haven't done it yet, set some environment variables to configure MCR. These are MCR install directories when it is installed by default. If you happen to have a custom MCR installation, make sure you are putting the correct paths:

echo 'export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/MATLAB/MATLAB_Compiler_Runtime/v83/runtime/glnxa64:/usr/local/MATLAB/MATLAB_Compiler_Runtime/v83/bin/glnxa64:/usr/local/MATLAB/MATLAB_Compiler_Runtime/v83/sys/os/glnxa64' >> ~/.bashrc 

echo 'export XAPPLRESDIR=$XAPPLRESDIR:/usr/local/MATLAB/MATLAB_Compiler_Runtime/v83/X11/app-defaults' >> ~/.bashrc 

echo 'export MCR_PATH=/usr/local/MATLAB/MATLAB_Compiler_Runtime/v83/extern/include' >> ~/.bashrc

echo 'export CPATH=$CPATH:$MCR_PATH' >> ~/.bashrc

• Source your .bashrc or close and reopen your current shell to confirm changes.
• Run the following to clone this repository:

git clone git://github.com/CorteXlab/gr-ephyl.git
cd gr-ephyl

• Now we have to make some modifications to CMakeLists.txt in order to include MCR and TurboFSK libraries when compiling:
• Open CMakeLists.txt and look for the 3 lines containing /home/othmane/ (They should be in lines 171, 172 and 179):

/home/othmane/opt/cx/opt/MATLAB_Compiler_Runtime/v83/extern/include
/home/othmane/opt/cx/include
...
/home/othmane/opt/cx/lib

• Replace these lines with the following:

${MCR_PATH}
${INSTALL_PATH}/include
...
${INSTALL_PATH}/lib

• Everything is set for our GNU Radio module to be compiled properly. Run the follwoing:

cp include/libTurboFSK_v4.h ${INSTALL_PATH}/include/
cp lib/libTurboFSK_v4.so ${INSTALL_PATH}/lib/
mkdir build
cd build
cmake -DCMAKE_INSTALL_PREFIX=${INSTALL_PATH} ..
make
make install

• Now generate the hierarchical blocks corresponding to the classes of nodes of the design:
  • Go to examples/, open all flowgraphs startins with hier_*.grc and generate their respective python files. Or copy-paste them from apps/ to $HOME/.grc_gnuradio.
  • Reload blocks

⁠How to use

Have a look at examples/, there are flowgraphs of loopbacks, disassembled hierarchical blocks and UHD compatible examples.

⁠Design

As mentioned before, there is two main classes/types of nodes:

• Base Station (BS) emulator (hier_bs.grc)
• IoT Sensor emulator, which is a network user in fact (hier_sensor*.grc)

⁠Documentation

• In progress