Semantic Communications
Semantic and Goal-Oriented AI Communications (SemCom)
A post-Shannon communication paradigm that transmits the underlying meaning and task-relevant information rather than raw bits.
Technical Explanation
Claude Shannon's classical 1948 information theory focused exclusively on reproducing exact bit sequences over noisy channels. Semantic Communications addresses Weaver's Level B (semantic) and Level C (effectiveness) communication. By training end-to-end Deep Neural Networks (Joint Source-Channel Coding - JSCC) across transmitter and receiver, 6G systems encode the essential semantics required to accomplish a specific task (e.g., classifying an obstacle or executing a remote robotic movement), yielding robust performance at signal-to-noise ratios far below classical Shannon limits.
Key Functions
- Joint Source-Channel Coding (JSCC) powered by deep neural networks
- Extraction and transmission of semantic features tailored to specific receiver tasks
- Extreme bandwidth reduction (>90%) for computer vision, robotics, and speech
- Unprecedented error resilience operating reliably in extreme low-SNR environments
- Goal-oriented optimization prioritizing data packets that impact real-world outcomes