Two-Way Semantic Communications Without Feedback

Semantic communications can significantly improve transmission efficiency, especially in the low signal-to-noise (SNR) regime. However, two-way semantic communications still remain an unexplored topic, which is a critical issue in many machine communication scenarios. Simply extending existing semantic communication systems to the two-way situation requires bidirectional information feedback during the training process, resulting in significant communication overhead. To fill this gap, we investigate a two-way semantic communication (TW-SC) system, where the information feedback can be omitted by exploiting the weight reciprocity in the transceiver. Particularly, the channel simulator and semantic transceiver are implemented on both TW-SC nodes and the channel distribution is modeled by a conditional generative adversarial network. Simulation results demonstrate that the proposed TW-SC system performs closing to the state-of-the-art one-way semantic communication systems but requiring no feedback between the transceiver in training process.