M. Sasaki, Naoki Shibuya, Kenichi Kawamura, Nobuaki Kuno, M. Inomata, W. Yamada, Takatsune Moriyama
{"title":"RNN Based Proactive Received Power Prediction Using Latest and Estimated Received Power","authors":"M. Sasaki, Naoki Shibuya, Kenichi Kawamura, Nobuaki Kuno, M. Inomata, W. Yamada, Takatsune Moriyama","doi":"10.1109/ISAP53582.2022.9998604","DOIUrl":null,"url":null,"abstract":"We report a method for proactively predicting received power using GRU (Gated Recurrent Unit), which is one of RNN (Recurrent Neural Network) as deep learning. One of the input data of the GRU is the latest received power obtained at the receiver, and another is the pre-estimated received power estimated at the future target position of the receiver. The training and validation data use received power of Wi-Fi measured in an indoor environment. According to the prediction method using the proposed model, the root mean squared error for the validation data is about 1.0 dB for the median received power. The prediction accuracy was improved by 1.7 dB compared with the baseline which uses the latest observed values.","PeriodicalId":137840,"journal":{"name":"2022 International Symposium on Antennas and Propagation (ISAP)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-10-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 International Symposium on Antennas and Propagation (ISAP)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ISAP53582.2022.9998604","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
We report a method for proactively predicting received power using GRU (Gated Recurrent Unit), which is one of RNN (Recurrent Neural Network) as deep learning. One of the input data of the GRU is the latest received power obtained at the receiver, and another is the pre-estimated received power estimated at the future target position of the receiver. The training and validation data use received power of Wi-Fi measured in an indoor environment. According to the prediction method using the proposed model, the root mean squared error for the validation data is about 1.0 dB for the median received power. The prediction accuracy was improved by 1.7 dB compared with the baseline which uses the latest observed values.