【摘要】 NCS雷达成像算法涉及到大规模的数据运算,为了满足雷达成像的实时性要求,将8K*4K矩阵数据进行任务划分,对算法流程进行并行化设计,映射到异构多核硬件系统并行计算,并在该原型演示系统上位机成图。通过对测试结果进行数据分析,得到基于原型多FPGA系统实现的NCS雷达成像算法具有实时性、高数据精度、高并行效率以及良好的硬件可扩展性等优点。在此基础上,初步设计了针对64K*32K大矩阵NCS算法实时成像系统的可行性方案。更多还原

【Abstract】 The NCS SAR algorithm involves large-scale computation.To meet the needs of real-time radar imaging,the 8k*4k source data is partitioned regularly,namely a reasonable parallelization design for NCS algorithm is worked out.Then the test software is mapped into heterogeneous multi-core system to be computed parallelly,and the ready-processed data is sent to the prototype demonstration system PC to form a mapping.The analysis of the test result shows the multi-core FPGA system which accomplishes NCS parallel design has the advantages of real-time,high data accuracy,high parallel efficiency and good hardware scalability etc.At last,a feasible scheme of real-time imaging system is provided aiming at the 64k*32k large matrix NCS algorithm.更多还原