srsRAN-matlab, developed by SRS, provides a number of MATLAB-based tools for testing and benchmarking the srsRAN Project software.
The project includes utilities for generating the test vectors used for testing in the srsRAN Project, MEX wrappers of a number of srsRAN PHY components, end-to-end simulators, and analysis tools for the baseband captures obtained with the srsRAN gNB (see phy_rx_symbols_filename
for how to obtain them).
For a better user experience, we suggest adding the root directory of srsRAN-matlab to the MATLAB search path
addpath path/to/srsran_matlab
For license details, see the LICENSE file.
srsRAN-matlab runs on MATLAB and builds upon the 5G Toolbox (tested on MATLAB R2022a, R2022b and R2023a under Linux, but other recent releases should also work).
The srsRAN Project is required to build the MEX wrappers and to run the applications that include them (see the MEX section for further details).
The development of srsRAN-matlab closely follows all new features of the srsRAN Project. For this reason, it is important that you always use the latest version of both software: this is the only way to ensure that test vectors agree with the srsRAN API and that MEX binaries compile. The following compatibility table provides a list of reference commits on both repositories that are guaranteed to work together.
srsRAN-matlab | srsRAN Project |
---|---|
0a235460 | 56a771df |
3c050654 | 0112729f |
73f47e3e | bcb449d7 |
f38c2c32 | 583c92b1 |
80f4a105 | 5e6f50a2 |
068e472a | e38e418b |
62c459a2 | 55c984b5 |
62c459a2 | dcd905cc |
040c50b5 | 0b2702cc |
842e0293 | 32dae89e |
b37e6ee5 | 50fe9623 |
b93615b7 | bcf941b3 |
85bb333e | 4d9f2232 |
6a268a15 | 2f90c8b6 |
fe585a5a | 1483bda3 |
1a81404d | 78238fd1 |
2a6f71b2 | f3ed07a5 |
0841ab86 | c33cacba |
52c07fbb | 40b17b42 |
f6a11714 | 4cf7513e |
a3bd2f8a | 4ac5300d |
d952d014 | 51e44a64 |
36fa859f | ee1d86cd |
36fa859f | e73b4618 |
e5155ae9 | 9d5dd742 |
latest | cda0033b |
The repository is organized as follows.
- Root directory: MATLAB classes for testing the srsRAN Project, see the Vector Tests section. This directory also contains the LICENSE, COPYRIGHT and README files.
+srsLib
: MATLAB implementation of several NR blocks and functionalities, mostly wrappers around MathWorks 5G Toolbox classes and functions. The internal structure of this directory follows the organization of the lib directory in the srsRAN Project. The content of this directory is intended for expert users only. Using functions and classes from this directory is discouraged.+srsMEX
: MEX version of a number of srsRAN blocks. The corresponding C++ source files are located in+srsMEX/source
. See the MEX section for more details.+srsTest
: Testing framework and utilities. The content of this directory is intended for expert users only. Using functions and classes from this directory is discouraged.apps
: End user applications, including simulators and analyzers. See the Apps section for more details.unitTests
: Repository CI/CD tools. The content of this directory is intended for expert users only.
Please read this file for a general overview of the project and its features.
All classes and functions are documented and extensive information can be obtained by typing help ClassName
at the MATLAB Command Window.
For support requests and community announcements, please use the srsRAN Project Discussion Board with the category "srsRAN-matlab".
The classes in the root folder can be used to generate test vectors (a header file and, typically, a tarball file containing the vectors) for the srsRAN Project (see the last paragraph of the Build Instructions).
Call runSRSRANUnittest
with the testvector
tag to generate the set of test vectors of all supported blocks with the command:
runSRSRANUnittest('all', 'testvector')
To generate the test vectors for a specific block, for instance pbch_encoder, simply run the command:
runSRSRANUnittest('pbch_encoder', 'testvector')
All generated files will be automatically placed in a directory named testvector_outputs. They can be copied to the proper subfolders inside the srsRAN Project folder with
srsTest.copySRStestvectors('testvector_outputs', 'path/to/srsRAN_Project')
The classes in the root directory define the tests following the MATLAB class-based testing framework. As a result, all the tools from the framework can be used to run and customize the tests. For instance, the following code will run the pdcch_candidates_common
tests only for a subset of values of the numCCEs
parameter.
% Select the block to test.
testedBlock = ?srsPDCCHCandidatesCommonUnittest;
% Test only for numCCEs = {24, 72}. By default, the test uses {24, 48, 72, 96, 120, 144}.
extParams = matlab.unittest.parameters.Parameter.fromData('numCCEs', {24, 72});
% Create a test suite.
vecTest = matlab.unittest.TestSuite.fromClass(testedBlock, 'ExternalParameters', extParams);
% Run the tests.
results = vecTest.run;
The directory +srsMEX
contains MEX versions of a number of classes and functions from the srsRAN Project. Typically, MEX can be accessed as class methods for a better user experience: for instance, the class srsMEX.phy.srsPUSCHDecoder
provides access to the MEX version of the srsRAN pusch_decoder
.
The following steps are needed to compile MEX binaries.
-
Export the srsRAN libraries: clone a local copy of the srsRAN Project (if not done already) and generate the CMake project with the
-DENABLE_EXPORT=True
option. This creates the filesrsran.cmake
in your srsRAN binary folder (that is, the folder at the top level of CMake build tree). -
Build the srsRAN Project: Follow the instructions in the srsRAN Project repository. You can use the target
srsran_exported_libs
to compile only the libraries needed by srsRAN-matlab and reduce compilation time. -
Generate the CMake project: In your local copy of srsRAN-matlab, do the following:
cd +srsMEX/source mkdir build cd build cmake ..
If the path to your
srsran.cmake
file matches the patterns~/srsRAN_Project/{build,build*,cmake-build-*}/srsran.cmake
,~/*/srsRAN_Project/{build,build*,cmake-build-*}/srsran.cmake
or~/*/*/srsRAN_Project/{build,build*,cmake-build-*}/srsran.cmake
, running CMake should find the exported libraries automatically. If this doesn't happen or if you have multiple copies of srsRAN on your machine, you should specify the path when running CMake.cmake -DSRSRAN_BINARY_DIR="~/new_srsran/new_build" ..
Similarly, you can use the CMake option
Matlab_ROOT_DIR
if you have multiple versions of MATLAB on your machine or if MATLAB is not in your system path.cmake -DMatlab_ROOT_DIR="/usr/local/MATLAB/R2023a" ..
-
Build the MEX: Simply run
make
to build the MEX binaries, which will be automatically placed in the proper folder to be accessed from thesrsMEX
library.Run
make doxygen
to build extra documentation, which can be accessed from+srsMEX/source/build/docs/html/index.html
.Depending on your setup, you may need to instruct MATLAB to use the system libraries instead of the internal ones: do the following and (re)start MATLAB.
cd /usr/local/MATLAB/R2023a/sys/os/glnxa64 sudo mv libstdc++.so.6 libstdc++.so.6.bak
The examples in this section assume you have MATLAB R2023a installed in the typical path
/usr/local/MATLAB/R2023a/
. For other MATLAB releases or paths, adapt the examples accordingly.
Call runSRSRANUnittest
with the testmex
tag to test the MEX. This command runs the same code as with the testvector
tag but sends the generated vectors directly to the MEX instead of writing them on file.
runSRSRANUnittest('all', 'testmex')
The folder apps
contains a number of applications and examples that use tools of the srsRAN-matlab project. Before running them, remember to add the main srsRAN-matlab folder to the MATLAB search path.
An instance of the PUSCHBLER class provides a simulator object for PUSCH BLER and throughput evaluation. The following example shows how to evaluate BLER and throughput at SNR = -6:0.2:-4
dB for the default configuration. For more information, enter help PUSCHBLER
at the MATLAB command line.
sim = PUSCHBLER % Create a PUSCHBLER object.
sim(-6:0.2:-4) % Run the simulation.
sim.ThroughputMATLAB % Display the evaluated throughput.
sim.plot % Plot the evaluated throughput and BLER vs SNR.
save my_sim.mat sim % Save the PUSCHBLER object, including the simulation results,
% to file my_sim.mat.
Function combinePUSCHSims
can be used to obtain a summary of several simulation results in graphic and table formats. For instance, the following command will draw the BLER and throughput curves from the PUSCHBLER objects saved in files my_sim1.mat
and my_sim2.mat
, as well as creating two tables, namely tableS
and tableM
, with the main simulation results using the SRS and MATLAB PUSCH decoder, respectively.
[tableS, tableM] = combinePUSCHSims(["my_sim1.mat", "my_sim2.mat"])
See help combinePUSCHSims
for more details.
An instance of the PUCCHBLER class provides a simulator object for the evaluation of the performance (in terms of BLER, detection and false detection probability, depending on the case) of the PUCCH processors, for PUCCH Formats 0, 1 and 2. The following example shows how to evaluate the PUCCH Format 2 BLER at SNR = -10:0
dB for the default configuration. For more information, enter help PUCCHBLER
at the MATLAB command line.
sim = PUCCHBLER % Create a PUCCHBLER object.
sim(-10:10) % Run the simulation.
sim.BlockErrorRateMATLAB % Display the evaluated BLER.
sim.plot % Plot the evaluated BLER.
save my_sim.mat sim % Save the PUCCHBLER object, including the simulation results,
% to file my_sim.mat.
An instance of the PRACHPERF class provides a simulator object for the evaluation of the PRACH probability of detection and of false alarm. The following example show how to evaluate the probability of PRACH detection at SNR = -6:0.2:-4
dB for the default configuration. For more information, enter help PRACHPERF
at the MATLAB command line.
sim = PRACHPERF % Create a PRACHPERF object.
sim(-26:0.5:-20) % Run the simulation.
sim.ProbabilityDetection % Display the evaluated detection probability.
sim.plot % Plot the evaluated detection probability.
save my_sim.mat sim % Save the PRACHPERF object, including the simulation results,
% to file my_sim.mat.
This app parses a section of the logs generated by the srsRAN gNB and returns carrier and channel (either PUSCH or PUCCH) configuration objects to be fed to one of the analyzers below (srsPUSCHAnalyzer or srsPUCCHAnalyzer). See the Configuration Parameters Section of the srsRAN Project documentation for information on how to configure the logging level of the SRS gNB to record the received samples.
See help srsParseLogs
for more details.
This app analyzes a PUSCH transmission from the baseband complex-valued samples corresponding to one slot, as received by the gNB. See the Configuration Parameters Section of the srsRAN Project documentation for information on how to configure the logging level of the SRS gNB to record the received samples.
See help srsPUSCHAnalyzer
for more details.
This app analyzes a PUCCH (Formats 1 or 2) transmission from the baseband complex-valued samples corresponding to one slot, as received by the gNB. See the Configuration Parameters Section of the srsRAN Project documentation for information on how to configure the logging level of the SRS gNB to record the received samples.
See help srsPUCCHAnalyzer
for more details.
This app analyzes a PRACH transmission from the baseband complex-valued samples corresponding to one PRACH occasion, as received by the gNB. See the Configuration Parameters Section of the srsRAN Project documentation for information on how to configure the logging level of the SRS gNB to record the received samples.
See help srsPRACHAnalyzer
for more details.
This app displays the content of a resource grid (all subcarriers and one slot) as a heat map of the resource element amplitudes. See the Configuration Parameters Section of the srsRAN Project documentation for information on how to configure the logging level of the SRS gNB to record the received samples.
See help srsResourceGridAnalyzer
for more details.
The class unitTests/CheckTests implements a series of checks to provide a basic level of quality assurance for the unit tests in the root folder.
These checks have been designed mainly for automatic CI/CD procedures. Nevertheless, they can be executed locally by running the following commands from the srsRAN-matlab root folder.
addpath .
runtests("unitTests/CheckTests.m")
The class unitTests/CheckSimulators carries out short runs of the simulators in the Apps folder to ensure their functioning.
These checks have been designed mainly for automatic CI/CD procedures. Nevertheless, they can be executed locally by running the following commands from the srsRAN-matlab root folder.
addpath .
runtests("unitTests/CheckSimulators.m")
The class unitTests/CheckAnalyzers carries out a demo run of the analyzers in the Apps folder to ensure their functioning.
These checks have been designed mainly for automatic CI/CD procedures. Nevertheless, they can be executed locally by running the following commands from the srsRAN-matlab root folder.
addpath .
runtests("unitTests/CheckAnalyzers.m")
The classes CheckPUSCHConformance, CheckPUCCHF0Conformance, CheckPUCCHF1Conformance and CheckPUCCHF2Conformace run a set of conformance tests (as defined in TS38.104 and TS38.141) of the corresponding PHY channel receivers.
These checks have been designed mainly for automatic CI/CD procedures. Nevertheless, they can be executed locally by running the following commands from the srsRAN-matlab root folder (be aware that these tests may run for several hours).
addpath .
runtests("unitTests/CheckPUSCHConformance.m")
runtests("unitTests/CheckPUCCHF0Conformance.m")
runtests("unitTests/CheckPUCCHF1Conformance.m")
runtests("unitTests/CheckPUCCHF2Conformance.m")