GNN-Based Decentralized Radio-Mapping Approach Introduced at ICASSP 2026
- 5월 14일
- 1분 분량
최종 수정일: 6월 3일
On May 4, during a special session at ICASSP 2026 in Barcelona, Spain, Ph.D. candidate Hongseok Jung presented a multi-agent decentralized radio-mapping technology in the Special session of Multi-Modal Signal Processing and AI for Communications and Sensing in 6G and Beyond.
To overcome the critical limitations of infrastructure-less and unstructured environments, the proposed approach leverages multipath propagation and Graph Neural Networks (GNNs). This approach enables collaborative environment mapping using only distance measurements from a single antenna, eliminating the need for centralized anchors.
▲ High-Precision Environment Mapping Results using the Proposed GAT Algorithm ▲
The presented work, titled "From Echoes to Maps: Collaborative Learning of Unstructured Spaces with Multipaths and Graph Neural Network," marks a significant advancement in next-generation wireless localization and Radio-SLAM.
▲ WSL member at ICASSP 2026 (Barcelona, Spain) ▲
댓글