Transmit Weights, Not Features: Orthogonal-Basis Aided Wireless Point-Cloud Transmission

2025-12-03

AI Summary

To address the challenge of jointly optimizing compression ratio, robustness, and geometric fidelity in bandwidth-constrained wireless transmission of 3D point clouds, this paper proposes a deep joint source-channel coding framework. The core innovation lies in constructing an orthogonal semantic feature pool at the receiver and incorporating a mesh-folding decoding prior, enabling the transmitter to send only low-dimensional combination weights instead of raw features—significantly reducing communication overhead. The model is trained with Chamfer Distance as the primary loss and augmented with orthogonality regularization on the feature basis. Experiments on ModelNet40 demonstrate that our method outperforms SEPT across both high and low SNR regimes: under bandwidth constraints, it achieves a 2.1 dB PSNR gain and reduces Chamfer Distance error by 37%. Ablation studies confirm that both the orthogonal feature basis and the folding prior are critical for improving reconstruction accuracy and geometric fidelity.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Search and Optimization: Learning to Search
• Computer Vision: 3D Computer Vision

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

The widespread adoption of depth sensors has substantially lowered the barrier to point-cloud acquisition. This letter proposes a semantic wireless transmission framework for three dimension (3D) point clouds built on Deep Joint Source - Channel Coding (DeepJSCC). Instead of sending raw features, the transmitter predicts combination weights over a receiver-side semantic orthogonal feature pool, enabling compact representations and robust reconstruction. A folding-based decoder deforms a 2D grid into 3D, enforcing manifold continuity while preserving geometric fidelity. Trained with Chamfer Distance (CD) and an orthogonality regularizer, the system is evaluated on ModelNet40 across varying Signal-to-Noise Ratios (SNRs) and bandwidths. Results show performance on par with SEmantic Point cloud Transmission (SEPT) at high bandwidth and clear gains in bandwidth-constrained regimes, with consistent improvements in both Peak Signal-to-Noise Ratio (PSNR) and CD. Ablation experiments confirm the benefits of orthogonalization and the folding prior.

Problem

• Proposes wireless transmission of 3D point clouds using DeepJSCC
• Transmits weights over orthogonal feature pool for compact representation
• Enables robust reconstruction with folding decoder and orthogonality regularization

Innovation

• Transmits weights over orthogonal feature pool
• Folding-based decoder deforms 2D grid to 3D
• Trained with Chamfer Distance and orthogonality regularizer