The Fraunhofer-Gesellschaft currently operates 76 institutes and research institutions throughout Germany and is the world’s leading applied research organization. Around 32 000 employees work with an annual research budget of 3.4 billion euros. The Fraunhofer Institute for Integrated Circuits IIS, located in Erlangen, is the largest institute of the Fraunhofer-Gesellschaft with more than 1 200 employees.
Conventional digital communication systems rely on Shannon’s separation theorem, which states that optimal performance can be achieved by independently designing source and channel coding components. However, this principle holds only asymptotically for infinitely long packets. In modern low-latency and short-packet communication scenarios, such as real-time multimedia transmission, this separation results in performance loss due to several factors:
- The optimality of the separation theorem breaks down for short block lengths.
- Real-world source coding does not yield ideally independent and identically distributed bitstreams.
To overcome these limitations, recent research explores generative AI-based transmission schemes. For instance, demonstrates the effectiveness of generative AI in communication tasks1, while proposes a variational autoencoder (VAE) architecture that jointly encodes and modulates images, outperforming state-of-the-art separate digital solutions2.
You are passionate about innovation and eager to tackle fundamental challenges in future communication systems? Then have a look at our offer!
