Abstract
Digital filters implemented on microprocessors operate under the constraints of quantisation of data, finite lengths of data words and microprocessor registers, limited arithmetical capability and speed. In view of this, digital filters with integer coefficients have been proposed. We present new techniques that take into consideration quantisation of filter coefficients while still retaining their integer coefficient property. We illustrate this with several applications to ECG signal processing. The paper presents design equations, pole/zero plots, frequency response and analysis of errors for quantised coefficient digital filters. Errors due to quantisation, roundoff, inaccurate coefficient representation and overflow are analysed. Finally, we present an example of a microprocessor-based system.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 18-25 |
| Number of pages | 8 |
| Journal | Medical & Biological Engineering & Computing |
| Volume | 25 |
| Issue number | 1 |
| DOIs | |
| State | Published - Jan 1987 |
Keywords
- Biomedical signals
- Digital filter
- ECG
- Errors
- Integer
- Microcomputer
- Quantisation
- Real time
ASJC Scopus subject areas
- Biomedical Engineering
- Computer Science Applications