面向VLSI的快速最小二乘自适应算法

该文章是自适应算法在实际系统中应用的一个尝试，它对基于QR分解的快速最小二乘自适应算法进行改进，打破了运算步骤之间的数据相关性，使算法能够由多个运算单元并发执行；同时，对运算形式进行了变形，使其适于由坐标旋转计算宏单元来完成．最后提出了整个系统的设计方案，并对其中的重要部分进行了详细讨论．系统设计中采用了一维拓扑结构，使得系统的硬件开销与系统阶数成线性关系，该系统具有很高的并行流水性和数据采样速率．

VLSI-oriented fast-least squares adaptive algorithm

An improved fast least_squares adaptive algorithm based on the QR decomposition is presented,It breaks down data relative to operation steps and executes the arithmetic parallel to many operation units. In addition, the operator format is transfigured to be finished by CORDIC macro unit. Finally, the design of an entire system is put forward,discussing in particular very large_scale integration(VLSI).An one_dimensional topology structure is adopted in the system design, which makes hardware spending linear with system rank. The system has very good level of parallel computation and fine pipeline and high speed of data sampling.