论文部分内容阅读
针对立方体钠卫星GNC信息处理系统高计算性能与低功率消耗相矛盾的问题,提出了一种资源限制型可重构并行信息处理方法。该方法采用紧耦合可重构并行信息处理架构,将GNC信息处理中需要多次迭代计算且不适合CPU处理的复杂软件算法,以动态部分重构硬件电路单元(DPR)的方式实现,采用基于互斥量的多核并行可重构资源调度算法,通过多核CPU并行管理与调度共享的DPR单元,完成软件算法的硬件加速与优化。实验结果表明,该方法实现了立方星GNC信息处理系统的高效实时快速处理,与传统信息处理方法相比,可节约50%左右的功耗,可应用于计算资源极为有限的星上信息处理领域,具有很好的工程应用前景。
Aiming at the contradiction between high computing performance and low power consumption of cubic sodium satellite GNC information processing system, a resource-constrained reconfigurable parallel information processing method is proposed. The method uses a tightly coupled reconfigurable parallel information processing architecture. The complex software algorithm that needs multiple iterative calculations in GNC information processing and is not suitable for CPU processing is implemented in a way of dynamically partially reconfiguring a hardware circuit unit (DPR) A mutex multi-core parallel reconfigurable resource scheduling algorithm, through the multi-core CPU parallel management and scheduling shared DPR unit to complete the hardware acceleration and optimization of the software algorithm. The experimental results show that this method can realize the fast and efficient processing of cubic GNC information processing system. Compared with traditional information processing methods, this method can save about 50% power consumption and can be applied to the field of on-line information processing with very limited computing resources , Has a good prospect of engineering applications.