Semantic Coded Transmission: Architecture, Methodology, and Challenges
share
Classical coded transmission schemes, which rely on probabilistic models and linear operations, have been always pursuing an elegant trade-off between effectiveness and reliability by source and channel coding techniques. However, a fundamental limit of these schemes is that they rarely take intelligent ingredients into consideration. In the future, communications toward intelligence and conciseness will predictably play a dominant role, and the proliferation of connected intelligent agents requires a radical rethinking of current coded transmission schemes to support the new communication morphology on the horizon. The recent concept of "semantic-driven" offers a promising research direction. Integrating semantic features in coded transmissions by nonlinear operations to achieve content-aware communications shows great potential for further breakthrough in effectiveness and reliability. Moreover, this form of coded transmission based on semantics is intuitively more consistent with essential demands of communications, i.e., conveying desired meanings or affecting conducts. This article aims to shed light on the emerging concept of semantic coded transmission (SCT), then present the general architecture and critical techniques of SCT, finally indicate some open issues on this topic.
READ FULL TEXT