Given finite samples of a stationary, perhaps noisy, non-Gaussian r-variate moving-average, MA(q) process, this study investigates cumulant-based identifiability conditions under which the MA coefficient matrices, the input statistics, and the order q, can be uniquely determined. The selection of a unique representative from the equivalence class corresponding to a given cumulant structure involves fewer restrictions than from the equivalence class corresponding to a given covariance structure. Two algorithms for estimating the (possibly) non-minimum-phase MA coefficient matrices are derived.
|Number of pages
|ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
|Published - 1988