For the past decade, Wi-Fi signal fingerprinting algorithm has been considered as a prevalent solution for indoor positioning systems. Fingerprinting based methods require a massive database of Wi-Fi signal samples to calibrate the indoor positioning system and to achieve a high location accuracy. Traditionally the calibration procedure requires human intervention and is very time-consuming, which makes a large-scale deployments of indoor positioning systems non-trivial. Objective of this research to minimise the manual workload by combining the conventional sampling algorithm with signal prediction. In contrast to traditional algorithms, proposed method requires only few signal samples to be collected and rest of the data are approximated using Discrete Fourier Transforms. The main objective of our research is to reduce the calibration effort while maintaining an acceptable location accuracy of the indoor positioning systems.