Benefits of Highly Synchronizing Huffman Encoding for Reducing Errors in Wireless Transmission of ECG Data
吉偉俊, William Gibson · 2013
When transmitting wireless ECG data from a heart monitor, it is useful to compress the data with a Huffman encoding so energy is saved and battery lifetime extended. However, single bit errors in Huffman can cause error propagation over multiple symbols. It is well known that different configurations of Huffman encoding can result in different amounts of error propagation. Therefore, using a highly synchronizing Huffman encoding for ECG transmission should result in more accurate data. This paper demonstrates through simulation that using Huffman codes which are highly synchronizing will result in higher fidelity ECG data than codes generated through a standard tree algorithm. In addition, using these codes produces no additional cost in transmission. The simulation is performed using a sample of existing ECG datasets from the MIT-BIH database [14] which are read and compressed using both standard and fixed-order [23] Huffman encodings. Each packet set is then subjected to a Gilbert-Elliot [7] [5] error channel to simulate transmission errors. Packets are then decoded and the waveform reconstructed. Result data is compared to the original data for its error content in terms of RMSE, R-Point detection, and two types of arrhythmia detection. Simulation shows that in every evaluation the reconstructed data from the fixed-order codes is superior to the standard. Further, the same degree of compression is achieved from both encodings. Results indicate that using highly synchronizing Huffman codes for compression of ECG data is a practical method to reduce error and comes at no additional transmission cost.