Speaker
Description
This talk introduces a GNU Radio Out-of-Tree (OOT) module that implements several blind source separation (BSS) algorithms, including Independent Component Analysis (ICA), Principal Component Analysis (PCA), Non-negative Matrix Factorization (NMF), and Adaptive Event Processing. The module integrates with GNU Radio’s flowgraph architecture, facilitating real-time signal processing within software-defined radio (SDR) applications.
We evaluate the effectiveness of these BSS techniques using both simulated datasets and real-world signals captured via SDR hardware. The primary metric for assessing separation quality is the Signal-to-Interference Ratio (SIR). Additionally, we compare the computational performance of the module across different platforms, analyzing runtimes on both CPU and GPU implementations.
The results highlight the varying performance of each BSS method depending on the nature of the input data and computational resources. Notably, GPU acceleration offers significant reductions in processing time, enhancing the feasibility of real-time BSS in complex signal environments. This OOT module extends GNU Radio’s capabilities, providing researchers and practitioners with a versatile tool for advanced signal separation tasks in SDR systems.
| Talk Length | 15 Minutes |
|---|---|
| Link to Open Source Code | https://github.com/gvanhoy/gr-bss |
