MODIFIED VERSIONS OF QRD-LSL ADAPTIVE ALGORITHM WITH LOWER COMPUTATIONAL COMPLEXITY

摘要

Three modified QR-decomposition-based least-squares lattice (QRD-LSL) algorithms are proposed. In the classical QRD-LSL, square-root operations are used, which are quite complex in a D.S.P implementation. The proposed algorithms reduce and even eliminate square-root operations and achieve a computational complexity of O(M), where M is the model order. The simulation results prove that these algorithms keep the fast rate of convergence and the numerical stability specific to the family of QRD-LS algorithms.