An Efficient Joint Channel and Impulsive Noise Estimation Method Based on Variational Autoencoders

In Orthogonal Frequency Division Multiplexing (OFDM) wireless communication systems, the joint estimation of channel and impulsive noise (IN) has become a research focus. The Sparse Bayesian Learning (SBL) algorithm is widely adopted for this task due to its effectiveness in handling sparse signals. But, it suffers from high computational complexity and time-consuming iterations. To solve these issues, this paper proposes a Joint Channel and Impulsive Noise Estimation method based on Variational Autoencoders (JCIE-VAE). First, a lightweight encoder network structure is designed, where the joint vector composed of the channel and IN is treated as the latent variable in the VAE for estimation. This practice effectively reduced the computational complexity. Secondly, a learnable sparse prior distribution is introduced into the VAE, which allows simultaneous updates of both the prior and posterior distributions of the latent variable during training. As a result, the iterative process inherent in the SBL algorithm is avoided. Experimental results demonstrate that the proposed JCIE-VAE method significantly reduces computation time. It improves estimation efficiency without compromising the bit error rate (BER).