RDJSCC

In this paper, we investigate the lossy access of correlated source in the scenario of distributed deep joint source-channel coding (DJSCC). Based on characteristic that the mismatched or low-correlated source can lead to a degradation of transmission performance. We propose a robust DJSCC (RDJSCC) scheme to maximize the advantages of distributed DJSCC in noisy environments. Unlike exiting methods, RDJSCC explores the trade-off between complementarity and consistency of the correlated source. Accordingly, novel cross-view information extraction (CVIE) mechanism and complementarity-consistency fusion (CCF) mechanism are developed in RDJSCC to upgrade distributed DJSCC. Information-theoretic analysis proves the rationality of exploring the above trade-off. Simulation results show that our proposed RDJSCC can effectively leverage the advantages of correlated sources even under severe fading conditions, leading to an improved reconstruction performance.