Libiq Library

Libiq is a modular and extensible library designed for the manipulation, visualization, and classification of I/Q (In-phase and Quadrature) samples.

It is structured into four main classes:

• Analyzer
  Provides tools for manipulating and analyzing time-series I/Q samples.

  • Supports reading binary/CSV files and extracting real, imaginary, or complex components.
  • Includes FFT and PSD methods for frequency-domain analysis.

• Plotter
  Enables real-time visualization of I/Q signals through various plot types:

  • Scatterplots:
    • Plot I vs Q to visualize signal characteristics.
    • Plot magnitude or phase over time to capture signal evolution.
  • Spectrograms: Compute and visualize time-varying frequency content using FFT. It supports custom window size and overlap settings for tuning resolution:
    • Smaller windows capture fine-grained variations.
    • Larger windows provide a broader overview of long-term signal behavior.

• Preprocessor
  Handles I/Q data preprocessing for CNN training pipelines.

  • Converts binary/CSV files into structured datasets.
  • Implements energy peak detection to isolate RFI signals and ensure model generalization.

• Classifier
  Contains methods to train and test a lightweight CNN model for RF signal classification. It uses real/imaginary parts, magnitude, and phase as input features.

Libiq has been successfully tested with python 3.9, 3.10, 3.11 and 3.12.

If you use the libiq library to develop your own works, please cite the following paper:

@inproceedings{olimpieri2025libiq,
  author    = {Olimpieri, Filippo and Giustini, Noemi and Lacava, Andrea and D’Oro, Salvatore and Melodia, Tommaso and Cuomo, Francesca},
  title     = {{LibIQ: Toward Real-Time Spectrum Classification in O-RAN dApps}},
  booktitle = {Proceedings of the IEEE Mediterranean Communication and Computer Networking Conference (MedComNet)},
  year      = {2025},
  address   = {Cagliari, Italy},
  organization = {IEEE}
}

arxiv url: https://arxiv.org/abs/2505.10537

Installation

There are two different ways to install Libiq

Package repository installation

The easiest way to install libiq is through PyPI. Simply run:

pip install libiq

It also offers optional dependencies, such as ydata-profiling and scienceplot, which can be installed as follows:

To enable reporting features:

pip install libiq[report]

To enable enhanced plotting styles:

pip install libiq[styles]

Or, to install all optional features at once:

pip install libiq[all]

To verify that the library has been installed correctly, a test script is provided in the docs directory under the name test_libiq.py.

You can run it using pytest:

pytest -v test_libiq.py

Build bash script

We provide a bash script that autmatically performs the steps described in Source installation.

To run it you simply need to execute

sudo ./build.sh

Source installation

Prerequisites

Install the basic tools required to build the libraries:

sudo apt install graphviz swig -y

Build and Install Dependencies

To work properly, Libiq needs FFTW.

The installation steps for this library starts with the download of the sources from the official site of FFTW, in particular we need fftw-3.3.10

mkdir libs
wget -O "libs/fftw-3.3.10.tar.gz" https://fftw.org/fftw-3.3.10.tar.gz
tar -xzf "libs/fftw-3.3.10.tar.gz" -C "libs/"
rm libs/fftw-3.3.10.tar.gz

Then we build and install

cd libs/fftw-3.3.10
./configure --enable-shared --with-pic --enable-threads
make -j$(nproc)
sudo make install
cd ../../
sudo ldconfig

Build the Libiq Python Package

This repository uses hatch for building the package.

hatch build

Install the Package

pip install dist/libiq-*.tar.gz

Then if you want to install the optional dependencies, do as in Package repository installation