I Can't Believe It's Not Real: CV-MuSeNet: Complex-Valued Multi-Signal Segmentation

--The increasing congestion of the radio frequency spectrum presents challenges for efficient spectrum utilization. Cognitive radio systems enable dynamic spectrum access with the aid of recent innovations in neural networks. However, traditional real-valued neural networks (RVNNs) face difficulties in low signal-to-noise ratio (SNR) environments, as they were not specifically developed to capture essential wireless signal properties such as phase and amplitude. This work presents C MuSeNet, a complex-valued multi-signal segmentation network for wideband spectrum sensing, to address these limitations. Extensive hyperparameter analysis shows that a naive conversion of existing RVNNs into their complex-valued counterparts is ineffective. Built on complex-valued neural networks (CVNNs) with a residual architecture, C MuSeNet introduces a complex-valued Fourier spectrum focal loss ( CFL) and a complex plane intersection over union ( CIoU) similarity metric to enhance training performance. Extensive evaluations on synthetic, indoor over-the-air, and real-world datasets show that CMuSeNet achieves an average accuracy of 98.98%-99.90%, Strikingly, CMuSeNet achieves the accuracy level of its RVNN counterpart in just two epochs, compared to the 27 epochs required for RVNN, while reducing training time by up to a 92.2% over the state of the art. The rapid growth of wireless communication technologies has congested the radio frequency (RF) spectrum, creating critical challenges for efficient utilization. Traditional fixed spectrum allocation methods are insufficient to meet the surging demand from connected devices [25]. Cognitive radio offers a promising solution by enabling dynamic access to underutilized frequency bands [1], [9], [10], [14]. Detecting and segmenting signals within wideband spectrum environments, referred to as spectrum segmentation (Figure 1), is a critical challenge in cognitive radio systems [2], [26].