Least-squares storage-channel identification

Pulse (dibit) and step (transition) responses for magnetic-storage channels are important for detection-circuitry design and for comparison of various media, heads, and other channel components. This paper presents a least-squares procedure that can be used to identify the dibit and transition responses from measurements of the read-head response to any known data sequence written on the medium. The method yields significantly higher-quality estimates for the dibit and step shapes than does determining these same characteristics by measuring the average response to isolated transition or by performing a Discrete Fourier Transform (DFT) on the response to a pseudorandom data pattern. The new method can be implemented off line but also can be made sufficiently efficient to be implemented with a microprocessor for use in self-optimizing (adaptive) channel detection circuitry.