Implementation of a Least-Squares Multi-Harmonic Fitting Algorithm in the Multicore Cell Processor

Least squares (LS) methods minimize the LS error between a specific data model and an acquired data set. This is the case of sine-fitting algorithms, which are used to determine the sine wave parameters that best fit an acquired signal. In the case of signals with multiple harmonics, one can use a multi-harmonic fitting algorithm resembling the sine-fitting algorithm to estimate the amplitudes and phases of the harmonics. This paper describes the implementation of a LS multi-harmonic fitting algorithm using the multicore Cell processor to determine the fundamental frequency of a multi-harmonic signal and then the harmonic amplitudes and phases.