Sayyed Shafivulla, A. Patel, Mohammed Zafar Ali Khan
{"title":"大规模MIMO线性检测器的并行实现","authors":"Sayyed Shafivulla, A. Patel, Mohammed Zafar Ali Khan","doi":"10.1109/CSNDSP54353.2022.9907964","DOIUrl":null,"url":null,"abstract":"Implementing the m-MIMO detectors and precoders requires a series of matrix-vector multiplications. These multiplications typically have iterative forms which are suitable for serial implementation. Serial implementations impose a significant delay into the system, impacting the system’s latency and making implementations a problem. This paper considers a recently proposed linear m-MIMO detector and presents an efficient parallel technique to compute the detector’s estimation vector. We implement the computation on the Nvidia Tesla T4 graphics processing unit (GPU) with compute unified device architecture (CUDA) application programming interface using Google colab. Numerical and implementation results are presented to quantity the speedup of the proposed parallel algorithm’s runtime as compared to existing parallel algorithms.","PeriodicalId":288069,"journal":{"name":"2022 13th International Symposium on Communication Systems, Networks and Digital Signal Processing (CSNDSP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Parallel Implementation of a Massive MIMO Linear Detector\",\"authors\":\"Sayyed Shafivulla, A. Patel, Mohammed Zafar Ali Khan\",\"doi\":\"10.1109/CSNDSP54353.2022.9907964\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Implementing the m-MIMO detectors and precoders requires a series of matrix-vector multiplications. These multiplications typically have iterative forms which are suitable for serial implementation. Serial implementations impose a significant delay into the system, impacting the system’s latency and making implementations a problem. This paper considers a recently proposed linear m-MIMO detector and presents an efficient parallel technique to compute the detector’s estimation vector. We implement the computation on the Nvidia Tesla T4 graphics processing unit (GPU) with compute unified device architecture (CUDA) application programming interface using Google colab. Numerical and implementation results are presented to quantity the speedup of the proposed parallel algorithm’s runtime as compared to existing parallel algorithms.\",\"PeriodicalId\":288069,\"journal\":{\"name\":\"2022 13th International Symposium on Communication Systems, Networks and Digital Signal Processing (CSNDSP)\",\"volume\":\"1 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2022-07-20\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2022 13th International Symposium on Communication Systems, Networks and Digital Signal Processing (CSNDSP)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/CSNDSP54353.2022.9907964\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 13th International Symposium on Communication Systems, Networks and Digital Signal Processing (CSNDSP)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CSNDSP54353.2022.9907964","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
摘要
实现m-MIMO检测器和预编码器需要一系列矩阵向量乘法。这些乘法通常具有适合串行实现的迭代形式。串行实现给系统带来了很大的延迟,影响了系统的延迟,使实现成为一个问题。本文考虑了最近提出的线性m-MIMO检测器,并提出了一种有效的并行技术来计算检测器的估计向量。我们在Nvidia Tesla T4图形处理单元(GPU)上使用Google colab的计算统一设备架构(CUDA)应用编程接口实现计算。数值和实现结果表明,所提并行算法的运行速度比现有并行算法有所提高。
Parallel Implementation of a Massive MIMO Linear Detector
Implementing the m-MIMO detectors and precoders requires a series of matrix-vector multiplications. These multiplications typically have iterative forms which are suitable for serial implementation. Serial implementations impose a significant delay into the system, impacting the system’s latency and making implementations a problem. This paper considers a recently proposed linear m-MIMO detector and presents an efficient parallel technique to compute the detector’s estimation vector. We implement the computation on the Nvidia Tesla T4 graphics processing unit (GPU) with compute unified device architecture (CUDA) application programming interface using Google colab. Numerical and implementation results are presented to quantity the speedup of the proposed parallel algorithm’s runtime as compared to existing parallel algorithms.