Wireless Image Semantic Cooperative Transmission in Distributed Edge Networks: An Information Disentanglement Method
Wei Wang, Donghong Cai, Zhiquan Liu, Zhiguo Ding, Pingzhi Fan · 2024
In this paper, we consider the wireless image semantic collaborative transmission for distributed edge networks, in which distributed edge devices extract features of the same target image from different viewpoints and transmit them to an edge server to perform reconstruction. Although edge device collaborative transmission overcomes the lack of individual sensing capability, it greatly increases the communication overhead. To address this challenge, we propose a distributed cooperative deep joint sourcechannel coding scheme based on information disentanglement, called Two-view-DC-DeepJSCC-D. Through curriculum learning, the proposed model learns the common and private information of the transmission signal to guarantee image recovery and clustering performance. The simulation results show that the proposed network has a notable improvement in the quality of reconstructed images compared with other state-of-the-art methods and exhibits greater robustness across various signal-tonoise ratios. Moreover, our model can effectively learn explanatory features, which can reduce transmission redundancy and improve clustering performance.