A simplification to the fast FIR-FFT filtering technique in the DSP interpolation process for band-limited signals

Abstract

Frequently, when great amounts of sampled data are manipulated, it is necessary to reduce them to a fraction that truly represents the data. Here two elements are important; neither is it desirable for the amount of data to be too extensive; nor is it desirable to lose key information. For the data reconstruction, it is often required a interpolation technique. This work proposes an optimization of the fast FIR filtering-based DSP interpolation technique. The resulting interpolator is a feasible, simplified version of hardware implementations, which is ideal for the reconstruction of band-limited signals with adequate sampling. In case studies included in this work, wind and temperature signals are used. The temperature signal has a suitable sampling rate and with the use of the interpolator it presents an error of 3.95% for a reduction of 99.61% of the data. The wind signal is very unpredictable and it does not have a suitable sampling rate. Hence, the interpolator does not improve the reconstruction of the signal in comparison to the averages. Typically, an average signal value is recorded for each 10 minute period, in which, the wind signal presents an error 8.6 times greater than the temperature signal. In order to reduce this error in the wind signal, it is recommended to increase the sampling rate and the quantization levels in measurement equipment.