Semantic Metrics for Non Real-Time Applications

In this work, we investigate the issue of semantics in non real-time communication scenarios, that is, without stringent information delivery constraints. While semantics has been widely used in the context of communications, the definition of the semantic metrics has been limited to the real-time applications. We thus revise the basic concepts of the timeliness and the accuracy and propose their formal definitions in the case of non real-time applications characterized by a possibility of non-causal signal reconstruction at the destination. In particular, we link the concept of accuracy to the canonical first-order auto-regressive model of the signal, while the timeliness is defined by the time difference to reference samples of the signal. We compare the proposed semantic metrics in the case of different buffer management strategies and using numerical examples we show that the optimality of the transmission strategy changes when the real-time or non real-time applications are considered.

Automated summary

In this study, we investigate the issue of semantics in non real-time communication scenarios and propose new definitions of timeliness and accuracy in non real-time applications. By comparing the semantic metrics under different buffer management strategies, we find that the optimality of transmission strategies differs between real-time and non real-time applications.