ABSTRACT
The demand for increased mobile phone subscribers require an efficient radio network planning that involves an accurate prediction of propagation loss. Although various empirical loss models are in use, they are suitable for a particular environment or a specific cell radius. The tuning of a model is a process in which the parameters of the theoretical propagation model are adjusted with the help of measured values obtained from the experimental data. In this process, several model parameters can be changed. The objective of tuning is to obtain the values of the predicted model parameters to have a closest match with minimal prediction error to the experimentally measured data. In this paper, we have introduced a tuned least absolute deviation (LAD)-based Walficsh/Ikegami (W/I) model using measured UMTS strength data carried out in GRA, Benin City, Nigeria. We found that the performance of the tuned W/I model is the best as root mean square error is lower compared to least square based regression technique. The tuned model with LAD method best fit into BS 2 and BS 3 with an RMSE of 5.41 as compared to least square tuning method which has an average RMSE of 16.68 with BS 2 and BS 3. This successfully validates our tuning methodology and suggests that the tuned model is more accurate for the specified environment.
KEYWORDS: Optimization, Propagation Model Tuning, least absolute deviation nregression method, prediction error, electromagnetic waves, environment.