High bit rate signal processing techniques and characterization for digital storage systems
In this dissertation we examine various signal processing techniques for increasing the achievable bit rate of a storage system, as well as the areal density. Throughout the thesis examples of the techniques are given for application to digital magnetic recording, although the techniques can be applied to other storage systems as well.
Chapter 1 contains an introduction to digital storage systems and communication channels. In Chapter 2 a quantization scheme is described which combines quantization with branch metric generation for a maximum likelihood detector. Combining these steps into one and restricting the branch metrics to consecutive integer values produces a fast algorithm. The number of quantization levels is kept small through the use of non-uniform thresholds in the quantizer.
In Chapter 3 the class of partial response (PR) polynomials that are interleavable is explored. Frequency-domain properties of the polynomials are presented, including the appearance of nulls in the spectrum at specified frequencies. Detector structures for the polynomials can utilize M parallel identical decoders each running at 1/M of the overall channel rate. A particular interleavable PR polynomial, PR(4 + $\alpha$), is examined in more detail in Chapter 4. DMR heads are discussed, and a good match between the DMR step response and that of PR(4 + $\alpha$) when $\alpha$ = 1 is demonstrated, using mean square error and noise enhancement as the criteria. This polynomial is also known as PR5.
In Chapter 5 the use of array heads is discussed. In addition to increasing the throughput of a recording system, array heads can be used to provide fine servo control signals. The theory of estimating the offtrack misregistration using array heads is presented, and simulation results confirming the theory are shown.
In Chapter 6 experimental measurements on a real-time PR4 channel with Viterbi detection are given. The probability of error is measured as the channel is degraded in various ways. The sensitivity of the PR4 channel to these degradations is thus determined.