Semantic-Based Integrated Sensing, Computing, Communication, and Control for Goal-Oriented Applications

The coming industrial internet of things (IIoT) era is anticipated to see the proliferations of goal-oriented applications in real-time wireless control systems. In such systems, a low processing latency is required to guarantee the timely transmission of control information. To achieve this goal, this paper proposes a new integrated sensing, computing, communication, and control $(\text{ISC}^{3})$ architecture, where semantic communications are adopted to make sensible semantic inference (SI) for the control information over time. In particular, we introduce a new criterion, i.e., mutual information (MI), for control performance evaluation by drawing from the field of wireless communications. We calculate the MI by designing a semantic feature extractor (SFE) module at the transmitter (Tx) to identify the semantic correlation among its sensed control information over time, thereby adjusting the frequency of its control information transmission. Meanwhile, a semantic feature reconstructor (SFR) module is employed at the receiver (Rx) to predict the current control information based on its previously received information if there is no control information transmission from the Tx. Finally, on-site experimental results are provided, showing that our proposed scheme can significantly reduce the communication overhead while improving the control performance significantly.