Error Control Coding: From Theory to Practice

Rating:
Summary: The book I've been looking for
Review: Coding - error control, as well as compression and encryption - is in your CD player, cell phone, internet link, and just about every electronic signal today. Despite its critical and growing importance, it has always been a specialist topic.

Sweeney just changed that. I'm no coding specialist. Still, I need a working knowledge of error control coding (ECC: detection, correction, or tolerance). This is the first book I know of that presents ECC at a non-specialist level. It still requires some college-level math: finite fields, probability, and a flexible attitude towards notation. It does not require the reader to hack through pages of theorems, though. Instead, Sweeney makes each point using worked examples.

The book covers a wide range of useful topics. It starts with the basics. For example, I now know what "coding gain" is. Next, Sweeney introduces convolutional or streaming codes. He not only shows the techniques, but shows the figures of merit for each code and how to compute them - knowledge I can use.

He brings the same clarity to the more common block codes, including cyclic, BCH, and Reed-Solomon codes. This discussion is followed by more analysis, showing at least a half-dozen ways to understand the strengths and weaknesses of any code. On the way through, he shows how the ECC interacts with the signal's analog encoding. We see how the "constellation" of a communication signal affects choice of finite field and ECC, and how that differs from codes suited to magnetic or other simpler representation.

The final sections discuss the elements that make up the most advanced codes in common use today: the turbo codes. By the time this discussion appears, the reader is already familiar with different ways to combine codes and to evaluate the combination. The reader also understands hard and soft Viterbi decoding, and when they are preferred over max-MAP techniques. Unlike older ECC books that seemed desperate for simple decoding circuits, Sweeney shows us the computation-intensive techiques of iterative decoding.

I find only two systematic faults in this book. First, the index is weak, and the table of contents isn't detailed enough to make up for that lack. Second, all of Sweeney's diagrams adopt the bizarre convention of introducing a data stream from the right, so that computation proceeds towards the left. This practice may make sense for trellis decoding. For data flow or circuit diagrams, however, it not only violates convention, it violates normal English left-to-right reading order. It was a real struggle sometimes to pull the proper sequences of events from a diagram, when the reading order was "backwards."

The book's strengths far outweigh those minor problems. As I said, I'm not a coding specialist. Until I see a better book, though, this is the one I'll recommend to anyone needing a basic literacy in ECC. If someone needs more than the basics, I'll still recommend this for its introductory content and bibliography.

Rating:
Summary: Much needed easy to understand book
Review: What a pleasue to finally see a book that deals with this complex and huge field in such a clean and easy to understand manner. Common confusing points are very well explained. Take for example, log-liklyhood ratio. Read the two page explanation with examples and never be confused again. Decoding algorithms such MAP (used for Turbo codes) are explained with examples. Even Finite field math is covered in exquisite simplicity. This is simply a must have book for those involved in coding. ...