Answers are grounded in this paper and cite their sources.
The evolution toward sixth-generation (6 G) networks is increasingly driven by the convergence of artificial intelligence, semantic communication, and integrated sensing capabilities, enabling networks to operate as sustainable, intelligent, context-aware, and adaptive systems. Rather than treating sensing and communication as isolated functions, emerging 6 G architectures increasingly rely on AI-native, cross-layer integration to jointly optimize perception, data delivery, and network control. Unlike existing surveys that treat AI-native networking, semantic communication, integrated sensing and communication (ISAC), edge intelligence, and programmable physical environments as largely separate research streams, this paper develops a network-architecture-oriented review centred on their cross-layer integration. The distinctive contribution of the review is to position semantic communication and ISAC not as isolated enabling technologies, but as mutually coupled functions within an AI-native 6 G system architecture in which sensing-derived context, semantic relevance, edge intelligence, and orchestration decisions interact across physical, control, user, and application planes. The review is based on a structured literature-selection process covering Web of Science and Scopus publications from 2019 to 2026, complemented by relevant standardization-oriented documents from ITU, 3GPP, and ETSI. The selected literature is classified according to architectural role, including semantic/task processing, sensing and communication integration, programmable physical-layer support, edge intelligence, AI-native orchestration, and standardization relevance. The analysis gives priority to architectural coupling mechanisms: how ISAC-generated environmental awareness supports semantic encoding and decoding; how semantic importance can guide radio-resource allocation, RIS configuration, and edge inference; and how AI-native orchestration may coordinate these functions under latency, energy, reliability, and standardization constraints. The paper further identifies open challenges related to semantic reliability, architectural scalability, orchestration overhead, standardization gaps, security, and energy sustainability, and outlines future research directions for deployable AI-native 6 G networks.
No takes yet. Share an insight, caveat, or question.
Ionel Bostan (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: