Existing biomedical wavelet based applications exceed the computational,
memory and consumption resources of low-complexity embedded systems. In order
to make such systems capable to use wavelet transforms, optimization and
implementation techniques are proposed. The Real Time QRS Detector and
?De-noising? Filter are developed and implemented in 16-bit fixed point
microcontroller achieving 800 Hz sampling rate, occupation of less than 500
bytes of data memory, 99.06% detection accuracy, and 1 mW power consumption.
By evaluation of the obtained results it is found that the proposed
techniques render negligible degradation in detection accuracy of -0.41% and
SNR of -2.8%, behind 2-4 times faster calculation, 2 times less memory usage
and 5% energy saving. The same approach can be applied with other signals
where the embedded implementation of wavelets can be beneficial.
Identifying and Exploiting Spatial Regularity in Data Memory References
