Concatenated Codes and Iterative Decoders for Magnetic Recording Channels

存储在磁盘和磁带驱动器上的数据的完整性必须使用错误控制码来保护，以防止错误引起的噪声和失真。在相当长的一段时间内，在编码领域中最令人兴奋的发现是所谓的turbo码，其涉及以便于迭代（“turbo”）解码的方式级联（级联）两个不同的码。 这种编码技术在简单的通信信道上实现了接近理论极限的性能，这与其发现之前的任何代码都不同。 本研究探讨磁记录装置中错误保护的涡轮码及相关码的设计，磁记录装置可模拟为（非常困难的）通信信道。 本研究的目标是设计高性能的磁记录级联码及其迭代译码器。并行和串行级联码都将被考虑。用于磁记录的差错控制码设计在过去集中于面向网格的码的设计，以最大化接收序列对之间的最小自由欧几里德距离。这是一个众所周知的难题，因为通过增加编码及其伴随的开销，码间干扰增加，妨碍了代码的预期益处。另一方面，级联码不仅能够通过增加开销（降低码率）来增加最小距离，而且能够通过交织器增益来实现增益。早期的工作表明，使用级联码和迭代解码可以实现前所未有的增益。

The integrity of the data stored on magnetic disk and tape drives must be protected from error-inducing noise and distortion using error control codes. The most exciting discovery in quite some time in the area of coding is the so-called turbo code which involves a concatenation (cascading) of two distinct codes in a manner which facilitates iterative ("turbo") decoding. This coding technique has enabled performance near theoretical limits on simple communications channels unlike any code before its discovery. This research investigates the design of turbo codes and related codes for error protection in magnetic recording devices, which may be modeled as a (very difficult) communication channel. The goals in this research involve the design of high performance concatenated codes and their iterative decoders for magnetic recording. Both parallel and serial concatenated codes will be considered. Error control code design for magnetic recording has focused in the past on the design of trellis-oriented codes to maximize the minimum free Euclidean distance between pairs of received sequences. This has been a notoriously difficult problem since, by adding coding and its concomitant overhead, intersymbol interference is increased, hampering the intended benefits of the code. Concatenated codes, on the other hand, are capable of achieving gains not just by increasing overhead (lowering code rate) to increase minimum distance, but via interleaver gain. Early work has shown that unprecendented gains are achievable with concatenated codes and iterative decoding.