Applications of iterative decoding to magnetic recording channels.

Abstract

Finally, Q-ary LDPC (Q-LDPC) codes are considered for MRCs. Belief propagation decoding for binary LDPC codes is extended to Q-LDPC codes and a reduced-complexity decoding algorithm for Q-LDPC codes is developed. Q-LDPC coded systems perform very well with random noise as well as with burst erasures. Simulations show that Q-LDPC systems outperform RS systems.

Secondly, binary low-density parity-check (LDPC) codes are proposed for MRCs. Random binary LDPC codes, finite-geometry LDPC codes and irregular LDPC codes are considered. With belief propagation decoding, LDPC systems are shown to have superior performance over current Reed-Solomon (RS) systems at the range possible for computer simulation. The issue of RS-LDPC concatenation is also addressed.

Three coding schemes are investigated for magnetic recording systems. Firstly, block turbo codes, including product codes and parallel block turbo codes, are considered on MRCs. Product codes with other types of component codes are briefly discussed.

Magnetic recoding channels (MRCs) are subject to noise contamination and error-correcting codes (ECCs) are used to keep the integrity of the data. Conventionally, hard decoding of the ECCs is performed. In this dissertation, systems using soft iterative decoding techniques are presented and their improved performance is established.