Distributed Images Transmission with Related Feature Assistance: A DeepJSCC Approach

With the rapid development of emerging applications such as the Internet of Things (IoT) and distributed visual perception, massive amounts of correlated image data require efficient transmission under constrained bandwidth and noisy channel conditions. Although Shannon’s separation theorem provides a theoretically optimal basis for independent source-channel design, end-to-end joint optimization methods demonstrate higher performance potential in finite block length scenarios. This paper addresses the distributed image transmission problem with source correlation by proposing a Deep Joint Source-Channel Coding (DeepJSCC)-based framework. The scheme introduces a correlation feature extraction module at the receiver to uncover similarities among multiple sources and assist image reconstruction. Experimental results demonstrate that this method significantly improves reconstruction quality across various signal-to-noise ratios (SNRs), particularly excelling under small bandwidth ratios.