Tuning Selection Impact on Kriging-Aided In-Building Path Loss Modeling

How do you know select enough tuning dataset from measurements to guarantee model prediction accuracy? Tuning datasets are often selected based on simple random sampling with predefined rates. Usually, these rates determined as a / xmlns:xlink="http://www.w3.org/1999/xlink">b , where % of the data goes training and remaining testing. But it is not clear what extent in order minimize estimation path loss errors. It is, thus, required analyze performance channel modeling by selecting—among all measurement samples—appropriate dataset. Using radio deterministic Ray Launching techniques collect reliable samples, this letter analyzes impact selection—expressed terms mean absolute error cost—on a novel Kriging-aided in-building measurement-based model.