Abstract
Technological advances in wearable sensing systems and data visualization mechanisms provide the opportunity of data driven immersive and real time mobile computing applications. The age of big data presents several challenges one of which is resource efficient collection of high resolution data. State of the art data compression schemes often do not rise up to the challenge. As an example, compression schemes for brain sensors often have unsatisfactory compression at the loss of valuable information. Compressive sensing (CS) is a paradigm shift in information theory, which utilized sparsity in some transformed domain to accurately recover signals sampled below the Nyquist rate. According to recent studies, CS techniques are often limited in their efficiency in accurately representing shape properties of signals. The problem is that for signal recovery CS optimizes objectives such as mean square error in replicating the time series. For signals with complex shape properties, such as heart signals, often such metrics are not adequate, since the error might be imposed on certain shape features, which are important for decision making. Signals with complex shape characteristics are often sparse in non-linear domain. Compression techniques that take advantage of non-linear sparsity, however, often fail to achieve a good compression ratio. In this chapter, we will critically analyze the performance and compression efficacy of different CS approaches and will also discuss the possibility of combining traditional compressive sensing techniques with non-linear compression mechanisms to reap the benefits of both approaches.
Original language | English (US) |
---|---|
Title of host publication | Compressive Sensing in Healthcare |
Publisher | Elsevier |
Pages | 65-88 |
Number of pages | 24 |
ISBN (Electronic) | 9780128212479 |
ISBN (Print) | 9780128212486 |
DOIs | |
State | Published - Jan 1 2020 |
Keywords
- Deployment in practice
- Model based compressive sensing
- Morphology preserving compressive sensing
- Physiological signals
ASJC Scopus subject areas
- General Engineering