Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep joint source-channel coding for wireless point cloud transmission

Cixiao Zhang, Mufan Liu, Wenjie Huang, Yin Xu, Yiling Xu, Dazhi He

TL;DR

This work tackles the challenge of wireless point-cloud transmission by proposing PCST, a semantic end-to-end system that jointly optimizes source and channel coding. It combines a progressive multiscale resampling network to extract latent semantic features with a Deep JSCC codec that allocates bandwidth adaptively based on feature entropy, while transmitting minimal side information for geometry recovery. The approach achieves superior reconstruction quality with substantial bandwidth savings (>50% at equal quality) and robust performance across AWGN and Rayleigh channels, addressing cliff effects observed in SSCC. Overall, PCST offers a scalable, robust solution for real-time, bandwidth-efficient point-cloud delivery suitable for metaverse and immersive applications.

Abstract

The growing demand for high-quality point cloud transmission over wireless networks presents significant challenges, primarily due to the large data sizes and the need for efficient encoding techniques. In response to these challenges, we introduce a novel system named Deep Point Cloud Semantic Transmission (PCST), designed for end-to-end wireless point cloud transmission. Our approach employs a progressive resampling framework using sparse convolution to project point cloud data into a semantic latent space. These semantic features are subsequently encoded through a deep joint source-channel (JSCC) encoder, generating the channel-input sequence. To enhance transmission efficiency, we use an adaptive entropy-based approach to assess the importance of each semantic feature, allowing transmission lengths to vary according to their predicted entropy. PCST is robust across diverse Signal-to-Noise Ratio (SNR) levels and supports an adjustable rate-distortion (RD) trade-off, ensuring flexible and efficient transmission. Experimental results indicate that PCST significantly outperforms traditional separate source-channel coding (SSCC) schemes, delivering superior reconstruction quality while achieving over a 50% reduction in bandwidth usage.

Deep joint source-channel coding for wireless point cloud transmission

TL;DR

This work tackles the challenge of wireless point-cloud transmission by proposing PCST, a semantic end-to-end system that jointly optimizes source and channel coding. It combines a progressive multiscale resampling network to extract latent semantic features with a Deep JSCC codec that allocates bandwidth adaptively based on feature entropy, while transmitting minimal side information for geometry recovery. The approach achieves superior reconstruction quality with substantial bandwidth savings (>50% at equal quality) and robust performance across AWGN and Rayleigh channels, addressing cliff effects observed in SSCC. Overall, PCST offers a scalable, robust solution for real-time, bandwidth-efficient point-cloud delivery suitable for metaverse and immersive applications.

Abstract

The growing demand for high-quality point cloud transmission over wireless networks presents significant challenges, primarily due to the large data sizes and the need for efficient encoding techniques. In response to these challenges, we introduce a novel system named Deep Point Cloud Semantic Transmission (PCST), designed for end-to-end wireless point cloud transmission. Our approach employs a progressive resampling framework using sparse convolution to project point cloud data into a semantic latent space. These semantic features are subsequently encoded through a deep joint source-channel (JSCC) encoder, generating the channel-input sequence. To enhance transmission efficiency, we use an adaptive entropy-based approach to assess the importance of each semantic feature, allowing transmission lengths to vary according to their predicted entropy. PCST is robust across diverse Signal-to-Noise Ratio (SNR) levels and supports an adjustable rate-distortion (RD) trade-off, ensuring flexible and efficient transmission. Experimental results indicate that PCST significantly outperforms traditional separate source-channel coding (SSCC) schemes, delivering superior reconstruction quality while achieving over a 50% reduction in bandwidth usage.
Paper Structure (11 sections, 4 equations, 3 figures, 1 table)

This paper contains 11 sections, 4 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. The proposed method
  3. Architecture
  4. Progressive Multiscale Resampling Network
  5. Entropy-guided Deep JSCC
  6. Optimization goal
  7. PCST overhead
  8. Experiment
  9. Implementation details
  10. Performance comparison
  11. Conclusion

Figures (3)

  • Figure 1: The PCST encoder has two main modules: the progressive multiscale resampling network$C_e$ for extracting latent features, and the switchable fully connected layer$f_e$ for encoding these features into variable-length symbols. The decoder mirrors the encoder’s structure.
  • Figure 2: Visualization of PCG reconstruction.
  • Figure 3: PCST performance vs. CBR on D1-PSNR for (a) AWGN channel and (b) Rayleigh channel at SNR = 10 dB; PCST performance vs. SNR under fixed CBR for (c) AWGN channel (CBR=0.015) and (d) Rayleigh channel (CBR=0.020).