arXiv:2606.12858v1 Announce Type: cross Abstract: Conventional communication systems, including both separation-based coding and learning-based joint source-channel coding (JSCC), are typically designed under Shannon's rate-distortion theory. However, relying on generic distortion metrics fails to capture complex human visual perception, often resulting in blurred or unrealistic reconstructions. In this paper, we propose Joint Source-Channel-Generation Coding (JSCGC), a generative communication paradigm that replaces the conventional decoder with a generative model at the receiver. The receive
Advances in generative AI models are enabling new approaches to data transmission that move beyond traditional distortion metrics toward semantic fidelity.
This concept could fundamentally change how data is transmitted by prioritizing human perception and generative reconstruction over raw data integrity, leading to more efficient and semantically rich communication.
Communication systems may shift from a 'perfect reconstruction' paradigm to a 'generative understanding' paradigm, where receivers infer and create content based on transmitted instructions rather than exact bitstreams.
- · Generative AI model developers
- · Telecommunications companies (adopting new standards)
- · Content creators (leveraging generative reconstruction)
- · Traditional communication hardware manufacturers
- · Shannon's rate-distortion theory purists
- · Data integrity focused industries (if not adapted)
Increased efficiency in wireless communication by reducing the data rate needed for perceptual quality.
New security challenges as generative models at the receiver could introduce artefacts or malicious content not present in the original signal.
The blurring of lines between transmitted information and locally generated content, leading to philosophical and legal questions around truth and origin.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI