In the 1940s, Claude Shannon developed the information theory focusing on quantifying the maximum data rate that can be supported by a communication channel. Guided by this fundamental work, the main theme of wireless system design up until the fifth generation (5G) was the data rate maximization. In Shannon's theory, the semantic aspect and meaning of messages were treated as largely irrelevant to communication. The classic theory started to reveal its limitations in the modern era of machine intelligence, consisting of the synergy between Internet-of-things (IoT) and artificial intelligence (AI). By broadening the scope of the classic communication-theoretic framework, in this article, we present a view of semantic communication (SemCom) and conveying meaning through the communication systems. We address three communication modalities: human-to-human (H2H), human-to-machine (H2M), and machine-to-machine (M2M) communications. The latter two represent the paradigm shift in communication and computing, and define the main theme of this article. H2M SemCom refers to semantic techniques for convey-

We consider the scenario of inference at the wireless edge, in which devices are connected to an edge server and ask the server to carry out remote classification, that is, classify data samples available at edge devices. This requires the edge devices to upload high-dimensional features of samples over resource-constrained wireless channels, which creates a communication bottleneck. The conventional feature pruning solution would require the device to have access to the inference model, which is not available in the current split inference scenario. To address this issue, we propose the progressive feature transmission (ProgressFTX) protocol, which minimizes the overhead by progressively transmitting features until a target confidence level is reached. A control policy is proposed to accelerate inference, comprising two key operations: importanceaware feature selection at the server and transmission-termination control. For the former, it is shown that selecting the most important features, characterized by the largest discriminant gains of the corresponding feature dimensions, achieves a suboptimal performance. For the latter, the proposed policy is shown to exhibit a threshold structure. Specifically, the transmission is stopped when the incremental uncertainty reduction by further feature transmission is outweighed by its communication cost. The indices of the selected features and transmission decision are fed back to the device in each slot. The control policy is first derived for the tractable case of linear classification, and then extended to the more complex case of classification using a convolutional neural network. Both Gaussian and fading channels are considered. Experimental results are obtained for both a statistical data model and a real dataset. It is shown that ProgressFTX can substantially reduce the communication latency compared to conventional feature pruning and random feature transmission strategies.

The next-generation wireless networks are envisioned to support large-scale sensing and distributed machine learning, thereby enabling new intelligent mobile applications. One common network operation will be the aggregation of distributed data (such as sensor observations or AI-model updates) for functional computation (e.g., averaging) so as to support large-scale sensing and distributed machine learning. An efficient solution for data aggregation, called over-the-air computation (AirComp), embeds functional computation into simultaneous access by many edge devices. Such schemes exploit the waveform superposition of a multi-access channel to allow an access point to receive a desired function of simultaneous signals. In this work, we aim at realizing AirComp in a two-cell multi-antenna system.To this end, a novel scheme of simultaneous signal-and-interference alignment (SIA) is proposed that builds on classic IA to manage interference for multi-cell AirComp. The principle of SIA is to divide the spatial channel space into two subspaces with equal dimensions: one for signal alignment required by AirComp and the other for inter-cell IA. As a result, the number of interference-free spatially multiplexed functional streams received by each AP is maximized (equal to half of the available spatial degrees-of-freedom). Furthermore, the number is independent of the population of devices in each cell.In addition, the extension to SIA for more than two cells is discussed.

No abstract

No abstract