Wireless Semantic Communications for Video Conferencing

Video conferencing has become a popular mode of meeting despite consuming considerable communication resources. Conventional video compression causes resolution reduction under limited bandwidth. Semantic (SVC) maintains high by transmitting some keypoints to represent the motions because background is almost static, and speakers do not change often. However, study on influence transmission errors limited. In this paper, an SVC network based keypoint established, which dramatically reduces resources while only losing detailed expressions. Transmission in lead changed expression, whereas those conventional methods directly destroy pixels. error detector, such as cyclic redundancy check, cannot reflect degree expression changes. To overcome issue, incremental hybrid automatic repeat-request framework for varying channels (SVC-HARQ) incorporating novel semantic detector developed. SVC-HARQ flexibility bit consumption achieves good performance. addition, SVC-channel state information (CSI) designed CSI feedback allocate enhance performance dramatically. Simulation shows that proposed wireless system can remarkably improve efficiency.