01138nas a2200133   4500000000100000008004100001260000900042100001700051700001600068700001500084700001900099245008800118520079800206       2021                            d  bIEEE1 aErdem Çakan1 aAral Şahin1 aMert Nakip1 aVolkan Rodoplu00aMulti-Layer Perceptron Decomposition Architecture for Mobile IoT Indoor Positioning3 a<p>We develop a Multi-Layer Perceptron (MLP) Decomposition<br />
architecture for mobile Internet Things (IoT) indoor<br />
positioning. We demonstrate the performance of our architecture<br />
on an indoor system that utilizes ultra-wideband (UWB) positioning.<br />
Our architecture outperforms the following benchmark<br />
processing techniques on the same data: MLP, Linear Regression,<br />
Ridge Regression, Support Vector Regression, and the Least<br />
Squares Method for indoor positioning. The results show that our<br />
architecture can significantly advance the positioning accuracy<br />
of indoor positioning systems and enable indoor applications<br />
such as navigation, proximity marketing, asset tracking, collision<br />
avoidance, and social distancing.</p>