6 Conclusion

This document has introduced four digital signal filtering methods and demonstrated their use on AVR devices. Each of the filtering methods has a spectrum of advantages - and disadvantages:

Median Filter

The median filter, often overlooked, is a rather simple but effective “signal smoothing” technique: It removes spikes and other glitches nicely and demands only a few CPU cycles. The downside is that the result could become too smooth and “rounded.”

IIR Filter

The Infinite Impulse Response filter offers several features from “classic” signal filtering - namely variants of high, low, and band-pass. The versatility and moderate consumption of CPU cycles make it an attractive tool.

FFT

The Fast Fourier Transform is not a classic filter per se, but it reveals a lot of information about the input signal. Hence, FFT is a mainstay of digital signal processing. From an 8-bit MCU perspective, FFT can demand a lot of resources in terms of memory and CPU cycles. It can provide substantial help in interpreting the input signal when applied thoughtfully.

Kalman Filter

Neither a classic filter, the Kalman filter has similarities with a controller: A noisy input signal can be used to determine an (output) parameter. The CPU consumption is moderate. The Kalman filter shines in settings where the actual value is expected to propagate continuously.