Decision-Oriented Communications: Application to Energy-Efficient Resource Allocation

摘要: In this paper, we introduce the problem of decision-oriented communications, that is, goal source is to send right amount information in order for intended destination execute a task. More specifically, restrict our attention how should quantize so can maximize utility function which represents task be executed only knowing quantized information. For example, functions under form $u\left(\boldsymbol{x};\ \boldsymbol{g}\right)$, $\boldsymbol{x}$ might represent decision terms using some radio resources and $\boldsymbol{g}$ system state observed through its version $Q(\boldsymbol{g})$. Both case where known it realizations, provide solutions determine such quantizer. We show approach applies energy-efficient power allocation. particular, seen quantizing very roughly perfectly suited sum-rate-type maximization, whereas energy-efficiency metrics are more sensitive imperfections.