Maria Kaselimi, A. J. Roumeliotis, A. Z. Papafragkakis, A. Panagopoulos, N. Doulamis
{"title":"卫星通信中使用深度学习技术的站点分集交换预测","authors":"Maria Kaselimi, A. J. Roumeliotis, A. Z. Papafragkakis, A. Panagopoulos, N. Doulamis","doi":"10.1109/ICASSPW59220.2023.10193159","DOIUrl":null,"url":null,"abstract":"An efficient deep learning (DL) architecture for switching prediction in site diversity schemes for Q band (39.402GHz) links is presented. The paper proposes the implementation of a DL detector (D) model in each station, that raises a flag when a rain event occurs, exploiting the benefits of transformer networks. When the event is detected, a DL regressor (R) model is triggered to predict future attenuation values for the specific event in each station. Both detector and regressor models consist of attention mechanisms that identify temporal dependencies between the input sequence elements. The experimental evaluation along with state of the art techniques indicate promising results.","PeriodicalId":158726,"journal":{"name":"2023 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Site Diversity Switching Prediction AT Q Band Using Deep Learning Techniques in Satellite Communications\",\"authors\":\"Maria Kaselimi, A. J. Roumeliotis, A. Z. Papafragkakis, A. Panagopoulos, N. Doulamis\",\"doi\":\"10.1109/ICASSPW59220.2023.10193159\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"An efficient deep learning (DL) architecture for switching prediction in site diversity schemes for Q band (39.402GHz) links is presented. The paper proposes the implementation of a DL detector (D) model in each station, that raises a flag when a rain event occurs, exploiting the benefits of transformer networks. When the event is detected, a DL regressor (R) model is triggered to predict future attenuation values for the specific event in each station. Both detector and regressor models consist of attention mechanisms that identify temporal dependencies between the input sequence elements. The experimental evaluation along with state of the art techniques indicate promising results.\",\"PeriodicalId\":158726,\"journal\":{\"name\":\"2023 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW)\",\"volume\":\"10 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2023-06-04\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2023 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ICASSPW59220.2023.10193159\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2023 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICASSPW59220.2023.10193159","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Site Diversity Switching Prediction AT Q Band Using Deep Learning Techniques in Satellite Communications
An efficient deep learning (DL) architecture for switching prediction in site diversity schemes for Q band (39.402GHz) links is presented. The paper proposes the implementation of a DL detector (D) model in each station, that raises a flag when a rain event occurs, exploiting the benefits of transformer networks. When the event is detected, a DL regressor (R) model is triggered to predict future attenuation values for the specific event in each station. Both detector and regressor models consist of attention mechanisms that identify temporal dependencies between the input sequence elements. The experimental evaluation along with state of the art techniques indicate promising results.