Joint SNR and Rician K-Factor Estimation Using Multimodal Network Over Mobile Fading Channels

Abstract

This paper proposes a novel joint signal-to-noise ratio (SNR) and Rician K-factor estimation scheme based on supervised multimodal learning. In the case of using machine learning to estimate the communication environment, achieving high accuracy requires a sufficient amount of training data. To solve this problem, we introduce a multimodal convolutional neural network (CNN) structure using different waveform formats. The proposed scheme obtains “feature diversity” by increasing the modalities from the same received signal, such as sequence data and spectrogram image. Especially with a limited dataset, training convergence is accelerated since different features can be extracted from each modality. Simulations demonstrate that the presented scheme achieves superior performance compared to conventional estimation methods.