Computational Complexity Upper Bounds For Fingerprint-Based Point-Of-Interest Recognition Algorithms

Thanks to the great proliferation of mobile devices within the Internet of Things (IoT) paradigm, Smart Buildings and Smart Cities are becoming hot topics. In such smart environments, user localization plays a central role. More specifically, Point-of-Interest (POI) recognition is one of the most attractive Location-Based-Service (LBS) applications. This paper provides computational complexity upper bounds for fingerprint-based POI recognition algorithms. We have considered five algorithms reported in a former work: LRACI and its extended variant ELRACI, BeaconPrint, PlaceSense, SensLoc and SAPFI. For each of them a close-form of their computational complexity has been derived, both for the training and the recognition phase. The results obtained in this works allow a comprehensive and fair comparison of some of the most well-known POI recognition algorithms.