OFDM-Based Digital Semantic Communication with Importance Awareness

Semantic communication (SemCom) has received considerable attention for itsability to reduce data transmission size while maintaining task performance.However, existing works mainly focus on analog SemCom with simple channelmodels, which may limit its practical application. To reduce this gap, wepropose an orthogonal frequency division multiplexing (OFDM)-based SemComsystem that is compatible with existing digital communication infrastructures.In the considered system, the extracted semantics is quantized by scalarquantizers, transformed into OFDM signal, and then transmitted over thefrequency-selective channel. Moreover, we propose a semantic importancemeasurement method to build the relationship between target task and semanticfeatures. Based on semantic importance, we formulate a sub-carrier and bitallocation problem to maximize communication performance. However, theoptimization objective function cannot be accurately characterized using amathematical expression due to the neural network-based semantic codec. Giventhe complex nature of the problem, we first propose a low-complexitysub-carrier allocation method that assigns sub-carriers with better channelconditions to more critical semantics. Then, we propose a deep reinforcementlearning-based bit allocation algorithm with dynamic action space. Simulationresults demonstrate that the proposed system achieves 9.7% and 28.7%performance gains compared to analog SemCom and conventional bit-basedcommunication systems, respectively.