Optimization and Deployment of Joint Source-Channel Coding for Enhance Future 6G Network Performance

Research on semantic communication has advanced because of deep learning in the construction for collaborative (joint) resource-channel coding (JSCC) project. Deep learning technology (DL)-based JSCC systems currently in practice are developed based upon a preset signal-to-noise ratio (SNR); though, such design is not intended in situations when SNR is changing. Thus, new semantic adaptive system with the semantic adaption modal selectioning (SAMS) modules is developed for semantic communication (SC). It is named collaborative (joint) resource-channel coding with adaptive models (AMJSCC). Both the real-time channel circumstances as well as the model's computing capabilities are taken into account by collaborative (joint) resource-channel encoding (JSCE) & collaborative (joint) resource-channel decoding (JSCD) models. In order to further increase the precision of information recovery, residual networks with various layers are also being studied. Simulation findings show that the model is flexible with the SNR as well as communication services, and it can achieve increased recovery similarity. Meanwhile, it uses less store space and communication resources than the most advanced deep JSCC techniques.