COMPARATIVE ANALYSIS OF ADVANCED STATISTICAL TECHNIQUES FOR OPTIMIZATION OF HYBRID MOBILE RADIO PATH LOSS MODEL

Path loss optimization is an important requirement in the design and implementation phase of mobile radio systems. The accuracy of the propagation model is improved by optimizing the model parameters to reflect the real time environment in a better manner. The paper aims at optimizing the proposed hybrid Walfisch–Ikegami path loss model for cellular signals in urban environment. The hybrid model is optimised with traditional method of Multiple Linear Regression and advanced statistical techniques of Ridge and Robust Regression. Although the techniques are used for other applications, evaluating the performance in the context of optimizing the mobile radio path loss model is a novel idea. The statistically developed optimized path loss models are validated by field strength measurements. The performances of the optimised models are evaluated in terms of path loss, error metrics and Goodness of Fit (GOF) tests. The Ridge optimized path loss model gives the least values of Prediction error (1.1701), RMSE (0.0827) and other error metrics. The Goodness of Fit tests additionally validates the efficient performance of Ridge optimized path loss model.