This work is part of the BAI4SOW («Inteligencia (Artificial) de Negocio para Flujos de Trabajo Sociales”) project, previously introduced in JCIS 2016, that deals with the study of the social workflows of consumers inside an open mall.
One key objective inside our sub-project is to localize and track the consumers by means of using the on board phone sensors (GPS in outdoors and WiFi, combined with other sensors, in indoors). This user localization represents the base of the activity recognition system that feeds the database. Using this
user positioning database and applying data mining processes the most frequent patterns of activity will be detected.
The objective of our proposal was to improve the resolution of fingerprint-based indoor WiFi localization systems without increasing the site survey effort. This way, WiFi indoor localization systems could be used to locate users in large environments (such as malls) reducing the effort of constructing the WiFi database. To do so, we proposed an approach, based on Support Vector Regression, to estimate the received signal strength at non-site-surveyed positions of
the environment. Experiments, performed in a real environment, showed that the number and distribution of the positions needed to train the system can be
reduced to almost half without significantly increasing the mean distance error.