Published Paper Details:

Approximation computing emerged as a key option for balancing the need for accuracy with the need to use as little energy as possible. Some programmes, like multimedia and signal processing, can handle errors and still process information correctly. This means that the accuracy level is lower than what is required at the circuit level, but the service quality is still good enough at the application level. Automatic recognition of R-peaks is the first and most important step that an electrocardiogram (ECG) signal must go through before it can be processed and analyzed. As its name suggests, the Haar discrete wavelet transform (HDWT) is a simple pre-processing filter that works great for finding electrocardiogram (ECG) R-peaks in embedded systems with very little power, like wrist tech. A study is used to show an approximate HDWT hardware architecture for ECG processing that is very good at using energy efficiently. Our best choice, which includes pruning in the roughly HDWT hardware design, only needs seven extra things to be added. In addition, this article looks at how to use a truncation method to make things use less energy. This is done by looking at how the signal-to-noise ratio changes over time and how that affects the ECG peak-detection programme in the end. According to the findings of this study, our HDWT approximation hardware architecture proposal is capable of bearing larger truncation levels than the HDWT that was first developed. Finally, our results show that using our HDWT matrix approximation idea along with pruning and the highest level of truncation still achieves an average R-peak recognition performance accuracy of 99.68%. This results in a decrease of approximately nine times the amount of energy that was previously consumed.