16-QAM-Detection-MIMO

DEPENDENCIES

Python 3.7.5

Numpy 1.17.2

matplotlib 3.1.1

CVXPY 1.0.25

INSTALLATION

Easiest way is to create a new Anaconda environment and install all necessary dependencies. If prefer not using Anaconda, just install all dependencies using plain pip, the Package Installer for Python.

ANACONDA

conda create -n mimo python=3.7 matplotlib numpy cvxpy

conda activate mimo

MOSEK SOLVER

Solver used for optimization is MOSEK. In order to use it, you will need to request a mosek trial license and follow the steps indicated there.

FILES

mimo.py - Looping through all SNRs specified for obtaining symbol errors using SDR and simple quantization, eigenvalue decomposition and randomization approximation techniques.

plot.py - Calculate Symbol error rates from mimo.py output log file and plot them.

optimization.py - Optimization module, it performs the SDR, minimizing the trace(W*A)

utils.py - Diverse utils necessary to perform the MIMO detection.

detection.py - Principal detection module.

remove_comments.py - Script to remove comments from files.

report.pdf - Slides explaining the work realized in this repository. Used for class presentation.

REMOVE COMMENTS

As the repository being a mere illustration of MIMO detection, it is full of comments to help grasp all the information.

In order to remove comments from files, just run the script remove_comments.py. By default is set to remove comments from detection.py, creating a new file called detection_no_comments.py.

To delete comments from other files, just change the last line of remove_comments.py:

remove_comments('detection.py', 'detection_no_comments.py')

• Change detection.py to the file you want to remove comments (specify path if is outside of root directory)

• Change detection_no_comments.py to your output-No-comments-file that will be generated.

TODO

• Add code license
• Add support for Maximum likelihood, zeroforcing and other detectors to compare results with SDR.
• Create an object Detector to choose between different detectors implemented.

REFERENCES

[1] Semidefinite Relaxation for Detection of 16-QAM Signaling in MIMO Channels -- -- A. Wiesel ; Y.C. Eldar ; S. Shamai

[2] MOSEK

[3] CVXPY

[4] Mathworks - Compute BER for a QAM system with AWGN

[5] Google Colab