Adaptive Rate-controlled Semantic Communication System for Wireless Image Transmission
Guolong Chen, Kechao Qi, Yun Jia, Nan Jiang · 2024
Semantic communication is considered an effective method for transmitting sufficient information under poor channel conditions. In order to fully leverage channel state information (CSI) and dynamically select output feature dimensions, we have proposed an adaptive rate control (ARC) wireless image transmission scheme called ARC-DJSCC. The ARC- DJSCC model consists of a j oint semantic encoding and decoding module and a semantic feature selection module. The j oint semantic encoding and decoding module adaptively generates semantic features of the image based on the input image and channel conditions. The semantic feature selection module dynamically selects feature dimensions of the transmitted semantic features according to the channel environment. The experimental results show that our proposed scheme achieves an effective semantic feature extraction and rate control strategy under adverse channel conditions, enhancing both the efficient use of channel bandwidth and the quality of reconstructed images.