Efficient realization structure of the a priori lattice-ladder recursive least squares algorithm

An efficient pipelined realization structure of the a priori lattice-ladder least squares multichannel algorithm is introduced. The structure is shown to exhibit great modularity and high parallelism. The architecture is evaluated for cost and data throughput delay assuming pipelined operations. It is demonstrated that the throughput rate of the structure is O(1) and the cost is O(l/sup 2/N), where l is the number of channels and N is the order of the filter. >