{"title":"Low-complexity Delay-Doppler Symbol DNN for OTFS Signal Detection","authors":"Ashwitha Naikoti, A. Chockalingam","doi":"10.1109/VTC2021-Spring51267.2021.9448630","DOIUrl":null,"url":null,"abstract":"In this paper, we consider the problem of low-complexity detection of orthogonal time frequency space (OTFS) modulation signals using deep neural networks (DNN). We consider a DNN architecture in which each symbol multiplexed in the delay-Doppler grid is associated with a separate DNN. The considered symbol-level DNN has fewer parameters to learn compared to a full DNN that takes into account all symbols in an OTFS frame jointly, and therefore has less complexity. Under the assumption of static multipath channel with i.i.d. Gaussian noise, our simulation results show that the performance of the symbol-DNN detection is quite close to that of the full-DNN detection as well as the maximum-likelihood (ML) detection. Further, when the noise model deviates from the standard i.i.d. Gaussian model (e.g., non-Gaussian noise with t-distribution), because of its ability to learn the distribution, the symbol-DNN detection is found to perform better than the ML detection. A similar performance advantage is observed in multiple-input multiple-output OTFS (MIMO-OTFS) where the noise across multiple received antennas are correlated.","PeriodicalId":194840,"journal":{"name":"2021 IEEE 93rd Vehicular Technology Conference (VTC2021-Spring)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"10","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 IEEE 93rd Vehicular Technology Conference (VTC2021-Spring)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/VTC2021-Spring51267.2021.9448630","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 10
Abstract
In this paper, we consider the problem of low-complexity detection of orthogonal time frequency space (OTFS) modulation signals using deep neural networks (DNN). We consider a DNN architecture in which each symbol multiplexed in the delay-Doppler grid is associated with a separate DNN. The considered symbol-level DNN has fewer parameters to learn compared to a full DNN that takes into account all symbols in an OTFS frame jointly, and therefore has less complexity. Under the assumption of static multipath channel with i.i.d. Gaussian noise, our simulation results show that the performance of the symbol-DNN detection is quite close to that of the full-DNN detection as well as the maximum-likelihood (ML) detection. Further, when the noise model deviates from the standard i.i.d. Gaussian model (e.g., non-Gaussian noise with t-distribution), because of its ability to learn the distribution, the symbol-DNN detection is found to perform better than the ML detection. A similar performance advantage is observed in multiple-input multiple-output OTFS (MIMO-OTFS) where the noise across multiple received antennas are correlated.