TR2024-070

Graph-Based Analog Joint Source Channel Coding for 3D Haptic Communication


    •  Fujihashi, T., Koike-Akino, T., Corcodel, R., "Graph-Based Analog Joint Source Channel Coding for 3D Haptic Communication", IEEE International Conference on Communications (ICC), June 2024.
      BibTeX TR2024-070 PDF
      • @inproceedings{Fujihashi2024jun,
      • author = {Fujihashi, Takuya and Koike-Akino, Toshiaki and Corcodel, Radu}},
      • title = {Graph-Based Analog Joint Source Channel Coding for 3D Haptic Communication},
      • booktitle = {IEEE International Conference on Communications (ICC)},
      • year = 2024,
      • month = jun,
      • url = {https://www.merl.com/publications/TR2024-070}
      • }
  • MERL Contacts:
  • Research Areas:

    Communications, Robotics, Signal Processing

Abstract:

Haptic communication will be a key technology for extended reality (XR), robotics, and remote manipulation. The deformation magnitude of 3D deformable objects is a key attribute for realizing fine-grained remote manipulation. The typical solutions for sending the deformation magnitudes over wireless channels are to perform digital compression and transmission considering the channel quality, i.e., digital source- channel coding. However, the key problems of the solutions are 1) still large traffic and 2) catastrophic quality degradation due to channel quality fluctuation. This paper proposes a graph- based analog joint source-channel coding for 3D haptic com- munication. Specifically, Graph Fourier Transform (GFT)-based energy compression efficiently removes the redundancy across deformation magnitudes. In addition, the integration of unequal error protection and analog modulation prevents catastrophic degradation and gradually improves the reconstruction quality according to the instantaneous channel quality. Evaluation results using the deformation magnitude of various 3D objects show that the proposed scheme prevents quality degradation due to channel quality fluctuations and provides accurate deformation magnitude for remote users.