Performance improvement of NN based RTLS by customization of NN structure - heuristic approach

Abstract

The purpose of this research is to improve performance of the Hybrid Scene Analysis -Neural Network indoor localization algorithm applied in Real-time Locating System, RTLS. A properly customized structure of Neural Network and training algorithms for specific operating environment will enhance the system's performance in terms of localization accuracy and precision. Due to nonlinearity and model complexity, a heuristic analysis is suitable to evaluate NN performance for different environmental conditions. Efficiency of the proposed customization of a Neural Network is verified by simulations and validated by physical experiments. This research also concerns the influence of size of Neural Network training set. The results prove that, better localization accuracy is with a NN system which is properly customized with respect to a training method, number of neurons and type of transfer function in the hidden layer and also type of transfer function in the output layer.

Physical layer authentication of off‐body channels by probabilistic neural networks

“…In Jachimczyk et al, 15 a conventional feed-forward ANN was used to compare current measurements of the RSS and the time of arrival (ToA) with RSS/ToA patterns. Performance of the described system was improved in Jachimczyk et al 16 In Sun and Lo, 17 authors developed a feed-forward ANN as a gait signal estimator. Acceleration in the inverted gravity direction, which was captured by sensors on a chest, played the role of training targets.…”

“…In Jachimczyk et al, a conventional feed‐forward ANN was used to compare current measurements of the RSS and the time of arrival (ToA) with RSS/ToA patterns. Performance of the described system was improved in Jachimczyk et al…”