Empirical Valuation of Multi-Parameters and RMSE-Based Tuning Approaches for the Basic and Extended Stanford University Interim (SUI) Propagation Models

Abstract

In this paper, the prediction performance evaluation of Stanford University Interim (SUI) Model and the extended SUI model are presented. More importantly, the effectiveness of two model tuning approaches, namely, RMSE-based tuning and multi-parameter tuning are assessed based on empirical pathloss data obtained for a suburban area in Uyo, Akwa Ibom state. Although the RMSE tuning is quite simple, the results showed that in some cases it does not minimize the prediction error to an acceptable level (6dB to 7dB) for practical applications. However, in the two models, the multi-parameter tuning effectively minimized the prediction error to an acceptable level with mean prediction error of about 0.00001 dB, RMSE that are less than 2.45 dB and prediction accuracies above 98.2%. On the other hand, the RMSE-tuned models have mean prediction error of above ± 1.5 dB, RMSE that above 8.8 dB and prediction accuracies less than 94.3%. In all, the SUI model performed better than the extended SUI.

Keywords

• Pathloss
• Model Tuning
• Empirical Pathloss Models
• Stanford University Interim (SUI) Model
• Multi-parameter tuning
• RMSE-based Tuning