Comparative Analysis and Calibration of Empirical Path loss models at 1800 and 2100 MHz frequency Bands for Hawassa City

Abstract

Due to recent years’ fast development of mobile communication technologies, the demand for high capacity network and excellent quality of service with thorough coverage has become essential. And the basic criteria in planning this type of network is the concept of Path loss. Path loss models estimates the average path loss a signal encounter at a certain distance from a transmitter. But, each kind of path loss propagation model intended to forecast path loss in a specific environment which might be wrong in another environment. Incorrect propagation model might cause low data throughput, poor coverage, poor quality of service, high investment cost and call drops which is clearly being observed in Hawassa City. Therefore, one of the root cause of these problems is the usage of un-calibrated path loss model during the network planning stage. This thesis presents a comparative study of five empirical path loss models according to the measured data collected from selected sites in Hawassa city at 1800 and 2100 MHz frequency bands. The preferred five models for investigation are COST- 231, ECC-33, Hata, SUI, and Ericsson model. A drive test approach was used for data collection and Nemo Handy and Nemo Outdoor were used as measuring a tools for the test. And the measured data were analyzed and extracted to suitable excel form for further analysis using ACTIX analyzer. Error measuring tools such as RMSE, Standard Deviation (SD), Mean Absolute Percentage Error (MAPE), and Mean Absolute Error (MAE) were used to evaluate the closeness of actual measured path loss with the path loss calculated from the listed empirical models in all area types. Then the best suited models were statistically calibrated using the Least Square algorithm to enhance the estimation accuracy. The result shows that, at 1800 Mhz band ECC-33 was the best fit model for Hawassa Urban areas with RMSE of 4.18 and 2.46 before and after calibration respectively whereas Hata was the finest for Sub-urban with RMSE of 7.86 and 5.18 before and after calibration respectively. And at 2100 Mhz band ECC-33 was once again the best fit model for Urban areas with RMSE of 6.15 and 5.28 before and after calibration respectively while Hata model found to be the best model for sub-urban areas with RMSE of 6.11 and 2.86 before and after calibration respectively. Thus, the optimized models are recommended for better deployment and gives an accurate path loss prediction.