一种基于GPU的实时软件接收机协同调度方法  

A GPU based coordinated scheduling method for real-time software receiver

作　　者：陈思业 CHEN Si-ye(Shanghai Institute of Microsystem and Information Technology,Chinese Academy of Sciences,Shanghai 200050,China;Shanghai Engineering Center For Microsatellites,Shanghai 201210,China;School of Information Science and Technology,ShanghaiTech University,Shanghai 201210,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院上海微系统与信息技术研究所,上海200050 [2]中国科学院微小卫星创新研究院,上海201210 [3]上海科技大学信息科学与技术学院,上海201210 [4]中国科学院大学,北京100049

出　　处：《电子设计工程》2018年第21期30-33,37,共5页Electronic Design Engineering

摘　　要：本文基于实现实时导航卫星软件接收机的目的,采用中央处理单元(CPU)和图形处理单元(GPU)的协同调度方法,利用GPU的多核并行处理性能,通过对导航卫星软件接收机的信号并行化处理,结合信号处理时负载任务的历史信息,提出了一种动态的调度方法,该方法根据估计时间信息在CPU和GPU之间选择合适的设备来执行任务。所提出的方法在CUDA平台上对软件接收机中的信号进行处理,得出基于时间估算的协同调度算法能满足实时软件接收机的需求的结论。Based on the realization of real-time navigation satellite software receiver,this paper uses the coordinated scheduling method of CPU and GPU to make use of the multicore parallel processing performance of GPU. By parallelizing the signal of the satellite software receiver,processing and combining with the historical information of load task in signal processing,a dynamic scheduling method is proposed. The method selects the appropriate device between CPU and GPU to perform the task according to the estimated time information. The proposed method processes signals in the software receiver on the CUDA platform and concludes that the coordinated scheduling method based on time estimation can meet the requirements of the real-time software receiver.

关 键 词：软件接收机 图形处理单元 协同调度 CUDA 

分 类 号：TN965.5[电子电信—信号与信息处理]