Channel modeling and target design for two-dimensional optical storage systems

In this paper, we first show that the symbol response of the two-dimensional (2-D) optical storage (TwoDOS) channel can be computed by a one-dimensional (1-D) Hankel transform, instead of the 2-D Fourier transform. This results in a computationally efficient approach for generating readback signals in the presence of pit-size variations. We also show how to design a 2-D minimum mean-square error (MMSE) equalizer for channels with these distortions. Second, we present a novel way to jointly design equalizer and target under linear constraints on the target, by transforming the 2-D target design problem into a 1-D form. Using a 2-D Viterbi detector, we investigated different target constraints. The results show that the newly proposed "2-D monic constraint" is a reasonable target constraint for a TwoDOS system.