Low Complexity LMMSE Channel Estimation on GPP

This paper proposes an improved linear minimum mean square error (LMMSE) algorithm of adaptive order determination by taking advantage of the structure characteristics of time domain least square (LS) channel estimation based on general-purpose processor (GPP), which provides the approximate estimation approach of max-time delay and noise power. In addition, this algorithm achieves a practical channel estimation formula which greatly reduces the complexity of the algorithm by decomposing the autocorrelation matrix into some sub-matrixes on the foundation of correlation bandwidth. Then comparisons are made between the simulation performance of improved LMMSE algorithm with those of other estimation methods for further analysis. Finally, a hardware implementation scheme of the proposed algorithm on GPP is given.