New Approach to Compression of Waveform Data

Huge amount of storage and high transmission bandwidth are necessary for applications dealing with wave form data. Hence, there have been many attempts to efficiently compress waveform data. This paper describes the use of double pass linear predictor in a two stage lossless data compression algorithm. A Linear Predictive Coding procedure is modified to allow its implementation with Huffman encoding. The use of double pass linear predictor removes the coherent components in the waveform data in a significant way. The second stage uses the Huffman encoding to further compress the residue of the first stage. The advantages of using Huffman encoding in the second stage are its simplicity of implementation, its effectiveness and its near-optimal performance in encoding Gaussian sequences. Our experimental results demonstrate that the proposed method achieves higher compression ratio and the recovered waveforms have a good signal quality.