Fadli Damara

For a radio frequency (RF) signal separation task, we propose two models operating directly on the time-domain waveform: a Transformer U-Net, a convolution-attention based model with an encoder-decoder architecture where self-attention blocks are inserted in the bottleneck to refine its representations, and a finetuned discriminative WaveNet model. The mixture of signal to separate is based on the ICASSP 2024 Signal Processing Grand Challenge on Data-Driven Signal Separation in Radio Spectrum. Compared to the baseline WaveNet architecture, we observed competitive performance with the Transformer U-Net and performance gains when finetuning the WaveNet model. The submissions achieved the 2nd rank in BER score and 3rd rank in MSE score.

The projected gradient descent (PGD) method has shown to be effective in recovering compressed signals described in a data-driven way by a generative model, i.e., a generator which has learned the data distribution. Further reconstruction improvements for such inverse problems can be achieved by conditioning the generator on the measurement. The boundary equilibrium generative adversarial network (BEGAN) implements an equilibrium based loss function and an auto-encoding discriminator to better balance the performance of the generator and the discriminator. In this work we investigate a network-based projected gradient descent (NPGD) algorithm for measurement-conditional generative models to solve the inverse problem much faster than regular PGD. We combine the NPGD with conditional GAN/BEGAN to evaluate their effectiveness in solving compressed sensing type problems. Our experiments on the MNIST and CelebA datasets show that the combination of measurement conditional model with NPGD works well in recovering the compressed signal while achieving similar or in some cases even better performance along with a much faster reconstruction. The achieved reconstruction speed-up in our experiments is up to 140-175.