Adaptive Tuning of the Log-distance Model for Optimal Predictive Modeling of Pathloss Over Irregular Terrains

Abstract

Recent technological development has facilitated the deployment of different Mobile Broadband Cellular Network Systems (MBCNS), such as Long Term Evolution (LTE) and 5G New Radio (NR), globally. This development aims at satisfying the ever-data-hungry multimedia applications to guarantee good quality of service for mobile subscribers. One way mobile subscribers can continuously access and enjoy the services provided by the MBCNS is to put in place a well-planned signal coverage area, wherein accurate path loss estimation is prioritized. Particularly, with the regular technological advancements and evolution of mobile communication systems, particularly fourth and fifth, the development of accurate précised path loss models has become more critical for robust planning and optimization purposes. In practice, the conventional log-distance path loss model is suitable for path loss predictive modeling and estimation in plain signal propagation environments. However, the model is not suitable for irregular terrains. In order to adapt this model for application in irregular terrains, this paper proposes a modified log distance path loss model with an adaptive polynomial term. The modified long-distance path loss model provides efficient irregular terrain signal loss estimation parameters. In order to boost the prediction accuracy of the proposed model, two regression optimization methods, non-linear least square and weighted least square, were employed with the Levenberg-Marquardt (LM) algorithm to determine its relevant parameters. In terms of the coefficient of correlation and percentage error, the modified log distance path loss model with the optimized parameters showed 40-60% improvement in accuracy over the standard log-distance model for the path loss prediction across the six different locations investigated. Furthermore, the validation of the proposed model has been provided in order to ascertain the level of its prediction accuracies in other locations. Overall, the modified log-distance model showed remarkable accuracy and efficiency when deployed in a related wireless propagation environment.