Coding and equalization for the PR1-equalized magnetic recording channel
[Thesis]
B. M. Zafer
W. E. Ryan
New Mexico State University
1997
154
Ph.D.
New Mexico State University
1997
Many significant improvements in the design of heads, media, and coding and signal processing have contributed to the steady growth of data rate and recording densities in magnetic recording systems. In this dissertation, we focus on new coding and equalization techniques that improve the performance of magnetic recording systems while maintaining manageable complexity. We examine several partial response schemes for magnetic recording channels at high densities. In particular, we study Class I partial response signaling (PR1) and compare its performance to the state-of-the-art schemes, class IV (PR4) and extended class IV (EPR4). Using a simple usd{8\over 9}(c=13)usd polarity switch DC-free code, we show that PR1 (2-state Viterbi detector) outperforms PR4 (4-state) and compares well with EPR4 (8-state), at a user density of usdS\sb{u}=2.5.usd This result motivates the design of more sophisticated DC-free codes. Thus, we design two new high-rate DC-free codes using the ACH state-splitting algorithm with parameters usd{8\over 9}(c=9)usd and usd{\over}(c=4).usd We compare the performance of these two codes with the usd{8\over 9}(c=13)usd and four other DC-free codes that have appeared in the literature. Our results demonstrate that the rate 16:18 code performs better than all other codes. In this study, we use a semi-analytic technique that includes the effect of imperfect equalization. Our study also includes a combined-trellis detector whose trellis is matched to the product of the PR1 trellis and the code constraint trellis. It is observed that the combined-trellis detector performs better than the two-state PR1 detector (matched only to the channel) by 1.8 dB for the 16:18 code. Our results also lead to the optimal recording density for each of the high-rate codes considered.