Partial Response Equalization

Supplementing their traditional view as a signalling or coding method, this paper elaborates a reception perspective of partial response techniques. It treats the data equalization problem that arises when an a priori selected partial response is secluded from a noisy dispersive communication channel and the equalization effort is exclusively directed to its remainder. For a given subdivision of the channel, and relative to a mean-square quality measure, it derives the optimum linear and decision-feedback equalizers, their performances and their sensitivity to gain variations. These results indicate that the linear equalizer greatly benefits from an appropriate selection of the partial response. In contrast, the performance of the optimum decision feedback equalizer turns out to be essentially independent of the applied subdivision, leaving only potential advantages in the domain of error propagation. The results are quantified for a class of digital magnetic recording channels, revealing a pre-eminent influence of the information density upon the relative merits of distinct partial response / equalizer combinations.