A cordic-based QR-RLS multichannel lattice filter

This paper presents an array-based algorithm for multichannel (multiple-input, single-output) lattice filtering. The filter is formed by an array of units that are adapted locally and in parallel using a set of recursions that closely match those for single-channel lattice filters. The design, based on a known step-by-step modular decomposition approach, allows for unequal filter lengths to be specified for different input channels. Individual units are updated using a square-root (QR) recursive-least squares (RLS) algorithm in array form that exhibits highly favorable numerical behavior. This type of filter may be mapped onto hardware by resorting to the CORDIC algorithm to implement Givens rotations. A procedure based on three CORDIC steps is proposed to handle complex data that arise in several applications of multichannel filtering. The algorithm is compared by simulation with plain RLS and another multichannel RLS lattice algorithm over randomly-generated channels, and shown to significantly outperform them under low-precision fixed-point arithmetic.