Comparison of Dynamic Weighted Centroid and Optimization Model to Predict Mobile Device Location

This research gives a comparison between two models to predict the location of mobile devices with cellular network signals: the dynamic weighted centroid model (DWC) and the generalized triangulation optimization model (GTO). Our previous work shows that both models give similar performances in Jakarta, a city with a high density of Base Transceiver Stations (BTS) and tall buildings. In this research, we apply and compare the two models to predict the position of mobile devices based on 200 Network Measurement Reports (NMR) collected in the Malang region, a medium city that is much smaller than Jakarta and has less dense BTS. Based on the statistical description of the results, we find that the GTO model gives a smaller minimum error, maximum error, mean error, and standard deviation compared to those of the DWC model. The cumulative distribution function (CDF) of the GTO model is around 99% of the prediction an error of less than 372.4 m better than that of the DWC model. This indicates that the GTO model outperforms the DWC model in predicting mobile device position