Description: Author: Robert G. Gallager | PDF
This is a definitive text written by quite possibly the greatest information theorist this country has produced, next to Claude Shannon. The author is personally responsible for some of the most central results in the field for the last forty years.
Gallager's book was the standard text in the field for much of the time it has been in print. In recent years, other textbooks, most notably the one by Cover and Thomas, have become more widely used in introductory classes in information theory. There may be several reasons for this. One may be the relative lack of motivating material in Gallager's text. The author assumes the reader is already convinced of the importance of the problems to be treated in the book. Cover's book, on the other hand, has greater appeal to the general scientist/engineer interested in the fascinating connections between information theory and various other areas of human inquiry. A second reason may be the relatively high level of intellectual rigor required of the reader. Gallager is a brilliant theorist, and his precision and meticulousness are apparent in the pages. The casual undergraduate or general reader may find some of the analysis intimidating and abstruse (although if he reads carefully, the reader always finds plenty of physical intuition to support theoretical arguments). A final reason is that some of the material (although certainly not the most important ones) in the book has become dated, and many exciting new developments in information theory, such as multi-user channels, simply never had a chance to make it in.
The above qualifications, however, should in no way obscure the greatness of the book. For anyone who is serious about information theory, this book is INDISPENSABLE. The relatively lack of breadth is more than made up for by the depth of the treatment. The precision that may intimidate an undergraduate is a godsend for a researcher frustrated with the sloppiness that occurs occasionally (or quite often) in other texts. And while the book may miss some new developments, its treatment of classic Shannon theory, particularly the noisy channel coding theorem, is unparalleled in its elegance and power. Moreover, the book is often the only recourse the researcher has when troubled with the subtleties of mind-numbing issues such as the continuous-time Gaussian channel. Finally, almost every one of the exercises in the book is a gem. There are any number of papers in information theory which cite Gallager's exercises as proofs!
The only true regret the reviewer has regarding the book is that Gallager has not written a new edition.
Link 1(PDF) Or Link 2 (DJVU)