STATISTICAL-DATA ANALYSIS OF MAGNETIC RECORDING NOISE MECHANISMS

Statistical data analysis using empirical eigenfunctions, known as the Karhunen-Loeve expansion, is applied to characterize noise mechanisms in magnetic recording. Given any original data set and hence its correlation (covariance) matrix, an empirical orthogonal set of eigenfunctions can be obtained. The original data set can be expressed as an orthonormal expansion of these eigenfunctions. This feature of the Karhunen-Loeve expansion can be used to study dominant noise profiles extracted from a large number of magnetization transition data. Two simple models of magnetization transitions are first utilized to investigate the validity of this expansion. Noises induced by transition center shifting (jitter), transition width fluctuation, amplitude modulation, and combined effects are respectively identified by the first several most important eigenfunctions in the expansion. Eigenfunction expansions of transition data obtained from experiments and numerical simulations are also obtained. The ease and versatility of the K-L method makes it a promising tool for isolating and probing dominant physical sources out of a seemingly large variety of output responses.