聯合訊源通道編碼技術用於無線近身網路之心電訊號壓縮

Abstract

在無線近身網路中，心電訊號壓縮是一項用來監測心臟相關疾病的重要技術。由於以往的心電壓縮技術著重於高壓縮率以及高速傳輸，在編碼端有比較高的複雜度，並不適用於無線近身網路。據此，本論文採用向量量化和索引指定方法來設計低複雜度的訊號壓縮器。此外，我們在解碼端考慮了通道效應，設計了迴旋編碼器提供資料保護並採用聯合訊源通道解碼的架構，搭配所提新的事前機率初始方法，不但達到良好的訊號重建品質，也提升了解碼速度。在MIT-BIH資料庫模擬結果，相較於其他心電傳輸技術，本論文所採用的架構能以較低複雜度之壓縮器，搭配迴旋碼達成良好重建品質並提高1.5倍收斂速度。

A wireless body area network (WBAN) for ECG compression is a promising approach for monitoring cardiac disease. Based on percentage root-mean-square difference (PRD) requirements, many ECG compression techniques have been introduced to achieve high compression ratio (CR) but may not be suitable in WBAN due to their higher encoding complexity. In this thesis, the joint source-channel coding scheme is proposed to not only adopt vector quantization and index assignment as the ECG compressor, but apply channel decoding technique to ensure robust ECG transmission. Note that the encoder parameters can be modified to obtain better reconstructing ECG signals at fixed CR. In addition, a new a priori knowledge initialization algorithm is presented to provide 1.5 times converge speed in contrast to original iterative source-channel decoding algorithm. After simulating with the MIT-BIH arrhythmia database, the results show that the proposed schemes with low encoding complexity has higher reconstruction quality than previous works.