Optimizing Load Balancing inRaytracing for Radio Frequency PropagationPredictive models developed for simulation systems

Abstract: This thesis presents two predictive models with the aim of pre-allocating memory efficiently in the Ericsson Radio Frequency Propagation Simulation System through the prediction model, so as to achieve the effect of load balancing. The proposed approaches utilize interpolation and the Gradient Boosting Decision Tree (GBDT) method from machine learning for prediction and  apply both methods in multiple test scenarios. A series of evaluation methods were conducted to evaluate the performance of the proposed methods and a comparison was made between the two methods. The results indicate that the machine learning method shows good predictive performance with a large amount of input data, while interpolation method has a good prediction accuracy in specific regions.