Fast Indoor Radio Propagation Prediction Using Deep-Learning Dataset

<p>We show a dataset&nbsp;composed by&nbsp;Radio Maps Estimation&nbsp;(RME) and Cells&nbsp;Maps Estimation (CME) for&nbsp;the 5GHz band WIFI in indoor scenarios: it&nbsp;has 60 indoor constructions plans and 1000 distributions initially for a training process and 20&nbsp;indoor constructions plans and 50 distributions aditionals for a test process&nbsp;of access points&nbsp;to even construction. These distributions are random and several WLAN&#39;s structures: 1 to 5 access points.</p> <p>The above explain that we got a total of 61000 RME&nbsp;and CME, this presents that is a model without interference between channels.</p> <p>Every coverage map have like maximum power delivered is <em>Pr = Pt = 26</em> dBm (according to data from <a href="https://www.cisco.com/c/en/us/products/wireless/catalyst-9100ax-access-points/index.html">current commercial equipment</a>) and like minimum power a value noise established in <em>Pr = KTB</em>, where K is the Boltzmann&#39;s constant, T the enviroment temperatura equal to 290&deg;K&nbsp;and&nbsp;B the band width equal to 80MHz.</p> <p>Dataset DeepFIRP is the result of a lot of simulations by a <a href="https://doi.org/10.5281/zenodo.7983595">own software developed in MATLAB</a> that work with&nbsp;the<a href="https://mentor.ieee.org/802.11/dcn/14/11-14-0882-04-00ax-tgax-channel-model-document.docx">&nbsp;IEEE 802.11ax&nbsp;channel model</a>.</p> <p>The pictures have a depth of 8 bits and size of <em>256pixels X&nbsp;256pixels</em> equivalents to indoor constructions of <em>20 X 20</em> m². These ones make reference to offices&#39;s spaces at&nbsp;general or classroom.&nbsp;</p> <p>A application to this dataset and the codes used for generate it&nbsp;is found <a href="https://github.com/johanflorez98/Fast-Indoor-Radio-Propagation-Prediction-Using-Deep-Learning">here</a>,&nbsp;where we implement a U-Net model for theRME and CME in indoor enviroments. This investigation contribute in novels methods for estimate by fast way coverage and cells maps using deep-learning in comparation with the conventional phisics methods like dominath-path model or ray-tracing. Whats allows save a lot of amount time in the WLANs&#39;s designs.</p>