Exploiting Channel Diversity to Improve BLE Range-Finding Accuracy

Accurate distance estimation techniques are one of the major requirements for 6G systems. Traditional techniques when using Bluetooth Low Energy (BLE) technology, are based on log-normal shadowing model. However, these models have proven to be incapable of providing the needed accuracy. This paper utilizes a novel method of distance estimation by using a Naïve Bayesian (NB) Classifier to fuse information present in packets from three different advertising channels and classify the distance. This method of distance estimation was tested on a custom dataset and found to have a Mean Absolute Error (MAE) of 0.0234 m. It was compared to the traditional log-normal shadowing curve estimation method and the NB method of fusing channel information was found to outperform it.