Consolidate viability and information theories for task-oriented communications: the case of remote power plants

Official URL: https://dx.doi.org/10.1109/EuCNC/6GSummit60053.2024.10597114

The next generation of cellular networks, 6G, is expected to offer a handful of exciting applications and services, including holographic communications, machine-to-machine communications, and data sensing from millions of devices. Whether these services come with a Gbps rate requirement or call for massive connectivity, they translate to the need for more spectral resources that are deemed to be scarce and costly. The communications resource should be wisely managed through value-driven approaches that eliminate waste and continuously enhance the communication process. These management principles align with the Task-Oriented Communications (TOC) philosophy. The aim is to allocate the minimum necessary communication resources according to the receiver (actor) goal. In the pursuit of the goal, the receiver may encounter irregularities and unforeseen events, rendering unclear how to build knowledge on the receiver's state and communicate accordingly. Our management approach integrates viability theory and transfer entropy to maintain the actor within a viable space, in contrast to the conventional approaches that help the actor to pass through singular states. By favoring a set of viable states aligned with the receiver (actor) goal and gradually reducing information exchange through knowledge accumulation, our method enhances flexibility and minimizes the risk of the receiver entering a non-viable state, thereby optimizing the communication resource as per the receiver goal. We discuss these theories in the context of TOC and examine their application in the plant process control case.