A Sensor Fusion and Machine Learning Approach for Predicting Signal Power Path Loss in Wireless Communications
When it comes to Wireless Communication systems and their optimization in different environments, estimation of SPPL for different terrain models becomes one of the most tedious problems for Radio Frequency engineers. Since every terrain has their complex terrain structures and contains micro-variations, they end up with an ambiguous SPPL via scattering and absorption. Also, modern SPPL prediction models are error free since they use predefined estimation parameters for classified terrain model. Sometimes, terrain-related estimation errors may cause over undesirable SPPL level which is much larger than 5% tolerance error. To avoid this problem, one can benefit from 3D map of the environment by using Light Detection and Ranging (LiDAR) and Artificial Neural Network which is one of the most common Machine Learning (ML) algorithm. This tools can be utilized to classify the objects and their structures which are the main reason for scattering and absorption. The fusion process of LiDAR and corresponding color classified satellite images can be fused to extract desired tree canopies. In this dissertation, Machine Learning and image processing techniques will be used to model SPPL for deployment of WCS.