摘要
提出了实时并行任务粒度划分的具体设计方法,采用了系统内通信迟延最小的约束条件进行多处理器上的任务分配,解决了高速实时信号处理中的数据通信瓶颈问题。在自行设计的超立方体结构的通用信号处理机上,实现了空时二维自适应信号处理和SAR实时成像,显著减小了数据通信迟延,提高了并行效率和性价比。
The task partition methods in real-time signal processing is presented, the constraints of minimizing communication delay in task allocation is proposed. The communication bottleneck in high speed real-time signal processing is resolved. The 2D space-time adaptive signal processing and real-time SAR imaging are implemented by a general purpose multiprocessor with hypercubic architecture. By means of all these proposed measures, the data communication delay can be decreased and the parallel efficiency and performance-price- radio is accordingly improved.
出处
《电路与系统学报》
CSCD
2004年第6期7-12,共6页
Journal of Circuits and Systems
基金
国防预研基金资助项目(41307010202)
总装跨行业基金资助项目(51407020201DZ0117)
关键词
实时处理
通信瓶颈
任务分配
real-time processing
communication bottleneck
task allocation