Abstract

The issues of timeliness and accuracy are becoming increasingly important with the advent of heterogeneous applications/networks and standardization of new services in areas of Cyber-Physical systems where the popular performance measures such as the throughput, error rate and packet loss lose their intuitive interpretation and new formal metrics must be defined on which the intuition will be built. To solve this problem, Semantic of Information has been recently introduced: the purpose and the final utility of the delivered data should be considered, and the performance is to be related to the semantics.

In this thesis, we explore two major issues in the area of semantic communications. First, we investigate the issue of wireless sensing in non-real time systems when the time delivery constraints are not restricted to be in real domain and the system tolerates delay in accessing the measurement results. To illustrate the application of this principle, we use the Markovian Gaussian models, and assuming the information semantic, we study the optimal estimation methods to non-causally reconstruct the source signal. We derive the explicit expressions and optimal buffer management policies for the proposed information accuracy metric.

Second, we evaluate the impact of the erroneous wireless control feedback channel using the Age of Information, one particular metric for semantic communications to capture the freshness and timeliness of information in real-time applications. To mitigate the impact of the imperfect feedback channel on the system performance, we adopt a Binary Asymmetric Channel model to control the detection accuracy of the control signals. We then compute the explicit expressions for the average Age of Information. Further, we show the optimum parameter design for the control channel model in order to minimize the average Age of Information.