Selection and Tuning Propagation Path Loss Model for Hawassa City, Ethiopia at 1800MHz Frequency

Path loss models estimate the average path loss a signal experiences at a particular distance from a transmitter. However, each type of existing path loss propagation model is designed to predict path loss in a particular environment that may be inaccurate in other different; hence selecting the best path loss model and optimizing it will minimize that inaccuracy. This work presents a comparative analysis of five empirical path loss models, COST- 231, ECC-33, Hata, SUI, and Ericsson model, with respect to the measured data from the 14 selected sites in Hawassa city, Ethiopia at 1800 MHz frequency bands. A drive test methodology was adopted for data collection and Nemo Handy and Nemo Outdoor were used as measuring tools for the test. Error measuring tools such as Root Mean Square Error, Mean Absolute Error, Standard Deviation, and Mean Absolute Percentage Error was used to select the terrain type of each site and the path loss model that best fits that site. The results showed not only Hawassa city consists of urban and Suburban terrains but also ECC-33 and Hata are better estimators for Hawassa Urban and Suburban areas with RMSE of 4.18 and 7.86 respectively. The model tuning using the Least Square Method reduced the RMSE of ECC-33 and Hata to 2.46 and 5.18 respectively. The reduction in RMSE shows that the tuned versions are close to the environment. Hence, using the tuned versions of the selected models will result in good cellular network design and enhance service quality.