A.V. Kobelev, S.I. Shchukin

Digital signal processing is a well-known part of electrical engineering and it is widely used in biomedical applications for increase signal-to-noise ratio, extract some signal-s features and so on. In conventional FIR filter design, too many coefficients needed to шьздуьуте high-quality filter (i.e. achieve narrow transition bands relative to sample rate). Most often such filters are impossible to implement due to finite-word-length effects (round-off noise and coefficient sensitivity). In the last years, digital signal processing has been further stimulated by multirate techniques. Multirate filters are based on conventional FIR (or IIR) digital filters and two specific operations: decimation and interpolation. The process of lowering the sample rate has been called decimation; similarly the process of raising the sample rate has been called interpolation. Both processed and their side-effects are discussed in the first part of this article. A key characteristic of multirate filtering is high computational efficiency, especially in case of breaking up a multirate system so that the processing occurs in a series of stages. Increased efficiency results primary from the relaxed filter design requirements on all stages. Discussion about choosing appropriate series of stage can be found in the second part of this article. The last part illustrates how the multirate digital techniques can be applied for high-quality low-pass filtration of breath pattern. We optimize not only computation efficiency, but also group delay time. As result, mulltirate filtration of breath pattern exceeds approaches existed (i.e. conventional linear filter and non-linear spline interpolation) in terms of efficiency and quality. In conclusion, multistage biosignal filtering makes it possible to simplify complexity of input analog circuits; get rid of phase correction; decrease power consumption and overall dimensions of biotechnical systems.