定点CORDIC算法的误差控制  

Error-control of Fixed-point Algorithm Based on CORDIC

作　　者：吴高奎[1] 严济鸿[1] 丁庆生[1] 

机构地区：[1]电子科技大学电子工程学院,成都611731

出　　处：《四川理工学院学报（自然科学版）》2010年第4期479-482,共4页Journal of Sichuan University of Science & Engineering(Natural Science Edition)

摘　　要：通过对CORD IC算法误差原因的分析,提出了一种降低定点算法误差累积的方法,从而可以使算法误差不随迭代深度增加而增加,进而得到更精确的函数值。文章首先分析了CORD IC算法的基本原理,依据此原理给出了典型CORD IC算法的基本结构,该结构可以用来有效计算超越函数的值;随即针对坐标转换时的误差累积效应以及误差较大的问题,给出了相应的误差分析以及修正的算法结构。FPGA仿真及实验结果表明,在不大幅牺牲速度的情况下,增加少量资源,可以在一定程度内控制算法的误差。After analysing the reason that the CORDIC algorithm error,this paper proposes a technique of fixed-point algorithm to reduce error-accumulate,thus the error of algorithm will not increase along with addition of the iterative depth.The basic theory of CORDIC algorithm is analyzed firstly,then the structure of typical CORDIC is presented,which can be used to compute transcendental function effectively,and later the analysis of error and modified-structure of the algorithm are given to the specific application of coordinate rotation.The results indicate that the technique of error-control given in this paper is feasible.

关 键 词：坐标旋转数字计算算法 误差控制 坐标转换 现场可编程门阵列 

分 类 号：TN4[电子电信—微电子学与固体电子学]