Mohamed Siala, ``Coding for The Magnetic Recording Channel,'' Ph. D., Ecole Nationale Supérieure des Télécommunications, Paris, France, 1994.
Advisor: Ghassan Kawas Kaleh
The first part of the dissertation deals with modulation and error-control
coding for the peak detection for the magnetic recording channel. We propose new methods for Run
Length Limited (RLL) coding based on several shaping techniques.
We show that shaping by structured vector quantization provides simple and
efficient codes with rates approaching asymptotically the Shannon capacity
of their constraints. We propose also a multi-level coding technique that
simplifies substantially the previous method. This technique enables us to
easily incorporate a multi-level error-control code. For this reason, we
are interested next in multi-level coding for the Lee metric. Therefore, we
propose a novel method based on the squaring construction to obtain good
codes for the Lee metric and very dense lattices for the 
distance.
Likewise, we propose an RLL coding system based on trellis shaping. We
suggest also a modification in order to include channel coding.
The second part is devoted to sampled detection systems. In particular, we propose new methods for simple and efficient coding for the dicode channel. First, our interest is focused on the determination of the cutoff density of the partial-response multi-track magnetic recording channel. Subsequently, we propose and study a matched spectral-null (MSN) code for the two-track dicode channel. This code provides an asymptotic coding gain of 3 dB in comparison to the uncoded dicode channel, while simplifying considerably the synchronization. In the sequel, we propose a class of block codes for the dicode channel whose decoder trellis structure is the same as that of the encoder, simplifying by the way the decoding process.
We study next several concatenated coding systems for the dicode channel. We present in particular an efficient and simple concatenated coding system where the inner code is the biphase code and the outer code is a punctured convolutional code. We also propose an iterative-decoding algorithm with soft outputs providing an important improvement in performance compared to the conventional decoding of serially concatenated codes. This algorithms can find its application in domains ranging from radio-mobile communication to satellite communication.
Author's address : Mohamed Siala, C/O Professor Ghassan Kawas Kaleh, Ecole Nationale Superieure des Telecommunications, 46, rue Barrault, 75 634 Paris CEDEX 13, France, Phone: 33 (1) 45 81 72 75, Fax: 33 (1) 45 89 00 20, E-mail: kaleh@com.enst.fr