- South Korea researchers created concretesc, a new way to improve wireless communication
- Concretesc avoids large code books, improves image transmission and reduces errors
- The team believes that it could help boost 6G networks, smart factories and medical care devices.
Researchers in South Korea have developed a new approach to semantic communication that could make future wireless systems faster and more efficient.
The new concretesc method was created by a team led by Dr. Dong Jin Ji, associate professor at the National University of Science and Technology of Seoul, and was published on June 19, 2025 in IEEE Wireless Communications Letters.
Semantic communication is a change in wireless technology where meaning is sent instead of unprocessed data. For example, by transmitting an image, the system prioritizes what the image represents instead of sending each pixel exactly. This saves band time and width and could be particularly useful for artificial intelligence and connected devices.
Potential for 6g
Existing systems often depend on the quantization of the vector, a process that uses “giant code books to store possible signal patterns. These code books are not only heavy to handle, but fight with errors and noise.
Concretesc resolves this with a different mathematical idea.
Despite its name, it has nothing to do with the material used in buildings. On the other hand, “concrete” here refers to a special probability distribution in automatic learning.
This tool allows to convert the continuous information into digital signals more smoothly, which allows the system to generate the bitstreams that it needs directly, without the burden of administering large code books.
“Unlike vector quantization (VQ), a state -of -the -art digitalization technique suffers from channel noise and code book during training, our frame offers a totally differentiable solution to quantization, allowing end -to -end training even under the noise of the channel,” Dr. Ji said.
“In particular, due to the nature of the concretesc that directly generates the required bit current, it is possible to train a couple of multiple feedback models with a relatively simple masking scheme,” he added.
In simulations, the researchers say that concretesc surpassed the methods based on VQ both in the structural similarity and in the maximum signal / noise ratio. It also reduced complexity, since its operations grow only with the length of bits instead of exponentially expanding with the size of the code book.
Researchers believe that this framework could play a role that is worthwhile in next -generation wireless systems, such as 6G.
They cite other potential uses, such as smart factories with ultra dense machine communications, as well as health and lifestyle monitoring systems fed by small devices enabled for AI.
