Resource efficient data compression algorithms for demanding, WSN based biomedical applications

J Biomed Inform. 2016 Feb;59:1-14. doi: 10.1016/j.jbi.2015.10.015. Epub 2015 Nov 7.


During the last few years, medical research areas of critical importance such as Epilepsy monitoring and study, increasingly utilize wireless sensor network technologies in order to achieve better understanding and significant breakthroughs. However, the limited memory and communication bandwidth offered by WSN platforms comprise a significant shortcoming to such demanding application scenarios. Although, data compression can mitigate such deficiencies there is a lack of objective and comprehensive evaluation of relative approaches and even more on specialized approaches targeting specific demanding applications. The research work presented in this paper focuses on implementing and offering an in-depth experimental study regarding prominent, already existing as well as novel proposed compression algorithms. All algorithms have been implemented in a common Matlab framework. A major contribution of this paper, that differentiates it from similar research efforts, is the employment of real world Electroencephalography (EEG) and Electrocardiography (ECG) datasets comprising the two most demanding Epilepsy modalities. Emphasis is put on WSN applications, thus the respective metrics focus on compression rate and execution latency for the selected datasets. The evaluation results reveal significant performance and behavioral characteristics of the algorithms related to their complexity and the relative negative effect on compression latency as opposed to the increased compression rate. It is noted that the proposed schemes managed to offer considerable advantage especially aiming to achieve the optimum tradeoff between compression rate-latency. Specifically, proposed algorithm managed to combine highly completive level of compression while ensuring minimum latency thus exhibiting real-time capabilities. Additionally, one of the proposed schemes is compared against state-of-the-art general-purpose compression algorithms also exhibiting considerable advantages as far as the compression rate is concerned.

Keywords: Biomedical signal processing; Data compression algorithms; ECG/EEG signal compression; Epilepsy monitoring; Performance evaluation; Wireless sensor networks.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Algorithms*
  • Data Compression / methods*
  • Databases, Factual
  • Electrocardiography
  • Electroencephalography
  • Epilepsy
  • Humans
  • Signal Processing, Computer-Assisted*
  • Telemedicine / methods*
  • Wireless Technology*