ORCAS - 分布式声学传感的正交表示压缩

中文摘要 AI 辅助

分布式声学传感（DAS）利用光纤电缆创建虚拟传感器阵列，以监测长距离的振动和应变。DAS在远程地震传感和基础设施保护中尤为有用，但会产生大量数据，带来处理挑战。本文提出了分布式声学传感的正交表示压缩（ORCAS），一个专为高效压缩DAS信号而设计的框架。ORCAS采用学习得到的正交变换进行快速稀疏编码，提供超过250 MB/s的吞吐量。它还包括一个针对DAS的编码方案，使用Lloyd-Max算法优化量化阈值以提高数据保真度。该系统实现了超过90 MB/s的实时吞吐量，并展现出优于现有标准（如JPEG2000和ZFP）的率失真性能。压缩比达到约13，SNR和PSNR值分别支持13 dB和28 dB。评估基于一个多样化的公开来源DAS数据集。

英文摘要

Distributed Acoustic Sensing (DAS) utilizes fiber-optic cables to create virtual sensor arrays that monitor vibrations and strain over long distances. DAS is particularly useful in remote seismic sensing and infrastructure protection but generates large data volumes that pose processing challenges. The paper presents Orthogonal Representation for Compression of distributed Acoustic Sensing (ORCAS), a framework designed for efficient compression of DAS signals. ORCAS employs a learned orthogonal transform for swift sparse encoding, delivering over 250 MB/s throughput. It also includes a DAS-specific coding scheme, optimizing quantization thresholds with the Lloyd-Max algorithm for improved data fidelity. The system achieves real-time throughput exceeding 90 MB/s and demonstrates superior rate-distortion performance compared to existing standards like JPEG2000 and ZFP. Compression ratios reach around 13, with SNR and PSNR values supporting 13 dB and 28 dB. Evaluations are based on a diverse, publicly sourced DAS dataset.

发表机构

• Eindhoven University of Technology（埃因霍温理工大学）

机构由 AI 辅助整理，请以论文原文为准。