Rate Adaptive Mechanism for Semantic Communication Systems: A Robustness Verification Approach

Traditional rate adaptive mechanism aims to maximize spectral efficiency by choosing the optimal transmission under the premise of perfect bit-level data transmission. However, in semantic communication systems, imperfect bit transmission could still lead to good performance due to the error correction capability of neural networks. In this paper, we propose a novel rate adaptive mechanism to maximize spectral efficiency while guaranteeing the performance of semantic tasks. In order to analyze the robustness of neural network inference, we introduce the robustness verification problem in semantic communications. Then, we propose and solve the modulation scheme selection problem subject to the robustness probability threshold constraint. Afterwards, a rate adaptive mechanism with the optimal modulation scheme is proposed. Simulations are conducted to validate the effectiveness of the proposed scheme, followed by some important open challenges for future related work.