arXiv:2602.04728v3 Announce Type: replace-cross Abstract: We propose a cross-attention Transformer for joint decoding of uplink OFDM signals received by multiple coordinated access points. A shared per-receiver encoder learns the time-frequency structure of each grid, and a token-wise cross-attention module fuses the receivers to produce soft log-likelihood ratios for a standard channel decoder without explicit channel estimates. Trained with a bit-metric objective, the model adapts its fusion to per-receiver reliability and remains robust under degraded links, strong frequency selectivity, an
The increasing complexity and dense deployment of wireless communication systems necessitate more efficient and robust signal processing techniques, coinciding with advancements in transformer architectures for diverse data types.
This development could significantly improve the performance and robustness of wireless communication networks, reducing the need for explicit channel estimation and enhancing uplink reception in challenging environments.
Wireless communication systems could achieve more reliable and faster data transfer with less computational overhead for channel estimation, potentially accelerating the deployment of advanced network technologies.
- · 5G/6G infrastructure providers
- · ML hardware manufacturers
- · Telecommunications companies
- · AI/ML researchers in networking
- · Traditional signal processing algorithm developers
- · Companies reliant on less efficient older OFDM systems
Enhanced uplink capacity and reliability for multi-AP OFDM systems.
Faster adoption and improved user experience for dense wireless networks, enabling more complex edge computing and IoT applications.
Reduced spectral efficiency bottlenecks in urban and industrial environments, potentially lowering infrastructure costs and enabling new high-bandwidth services.
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.LG