Scholarlink

<p>The increasing adoption of autonomous electric vehicles (AEVs) highlights the need for efficient visual data transmission in resource-constrained environments. This study proposes an optimized CNN-based autoencoder framework specifically designed for image compression and reconstruction over LoRa networks, which offer long-range communication with limited bandwidth. By integrating advanced compression techniques such as Huffman Encoding, Zlib Compression, and Adaptive Quantization, the framework effectively mitigates LoRa’s bandwidth limitations while maintaining high reconstruction quality. Experimental results on 80 × 80 pixel images demonstrate the system’s ability to balance transmission efficiency and image fidelity, achieving an average Peak signal-to-noise ratio (PSNR) of over 24&#xa0;dB. The model incorporates multiple loss functions, including Mean squared error (MSE) and Structural similarity index measure (SSIM), to enhance reconstruction performance. Its flexible architecture supports varying compression ratios and latent space sizes, enabling adaptability across different application scenarios. This research represents a significant step towards real-time image transmission for AEVs in low-bandwidth settings, paving the way for future advancements in scalable and robust data communication systems for autonomous vehicles.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimized CNN-Based Autoencoder for Efficient Image Compression in LoRa Networks for Autonomous Electric Vehicles

  • 期刊论文

  • 发表日期：

    2025年4月18日

查看全文
  • Reza Nurfaudzan Ash Shiddiq,
  • Arief Suryadi Satyawan,
  • Galura Muhammad Suranegara,
  • Muhammad Iqbal Fauzan

摘要

The increasing adoption of autonomous electric vehicles (AEVs) highlights the need for efficient visual data transmission in resource-constrained environments. This study proposes an optimized CNN-based autoencoder framework specifically designed for image compression and reconstruction over LoRa networks, which offer long-range communication with limited bandwidth. By integrating advanced compression techniques such as Huffman Encoding, Zlib Compression, and Adaptive Quantization, the framework effectively mitigates LoRa’s bandwidth limitations while maintaining high reconstruction quality. Experimental results on 80 × 80 pixel images demonstrate the system’s ability to balance transmission efficiency and image fidelity, achieving an average Peak signal-to-noise ratio (PSNR) of over 24 dB. The model incorporates multiple loss functions, including Mean squared error (MSE) and Structural similarity index measure (SSIM), to enhance reconstruction performance. Its flexible architecture supports varying compression ratios and latent space sizes, enabling adaptability across different application scenarios. This research represents a significant step towards real-time image transmission for AEVs in low-bandwidth settings, paving the way for future advancements in scalable and robust data communication systems for autonomous vehicles.