 |
Book Review
Statistical and Adaptive Signal Processing, by Dimitris Manolakis, Vinay Ingle, and Stephen Kogon, Artech House 2005, ISBN 1-58053-610-7
Reviewed by Vladimir Botchev [vladimir.botchev@analog.com]
This book is a re-issue of the highly successful McGraw Hill title, published a few years ago. It has been used as a basic book in highly respected IEEE courses on Adaptive Antenna Arrays. It is of the same caliber as the DSP practitioner’s must-have: Proakis & Manolakis, Digital Signal Processing: Principles, Algorithms and Applications (3rd Edition). It provides a solid background for further specialized study in any of the disciplines considered inside.
The book has 12 chapters, five appendixes and a website for Matlab code—very useful for performing experiments based on the book’s teachings. At one time, an extensive solution manual was also available to qualified instructors.
In the introductory Chapter One, the book considers its four major topics—spectral estimation, array processing, adaptive filtering and signal modeling, assuming almost zero knowledge of these topics. Chapter Two reviews fundamental discrete-time signal-processing topics, while making it all but unnecessary for the reader to consult outside sources in cases where a basic discrete time operation is needed. Chapter Three’s figures and visual aids make it one of the best introductions to random variables and stochastic processes in the engineering literature. Chapter Four, on signal modeling, introduces basic parametric models: all-pole, all-zero, and pole-zero—as well as the cepstrum of pole-zero models. Classical non-parametric (i.e., autocorrelation- or Fourier-transform-based) spectral estimation is detailed in Chapter Five.
A very thorough treatment is given to optimal filtration and prediction topics in the next three chapters: optimal filtration, introducing linear prediction; algorithms for optimal filters, such as Levinson-Durbin and other widely used matrix-inversion algorithms and implementations; and least-squares filtering and algorithms, such as orthogonalization and SVD (singular-value decomposition).
The next few chapters dig deeper, making good use of the material presented so far. Chapter 9 details parametric spectral estimation—where prediction plays a non-negligible role—along with minimum-variance and harmonic (e.g., Pisarenko) methods. Chapter 10, on adaptive filtering, is almost a small book by itself. All major algorithms are detailed, and—here especially—Matlab experiments are quite revealing.
Array processing is considered in Chapter 11. Beamformers based on both conventional delay-sum and optimum-signal-processing are considered (generalized sidelobe canceller). Also some details are given for MVDR (minimum-variance distortionless response) beamforming (essentially the same technique as in spectrum modeling), which is a preferred technique for many microphone-array applications. The final chapter covers what may be considered as advanced topics, such as polyspectra, blind deconvolution, and self-similar models.
The appendixes are primarily concerned with intricate details of vector and matrix algebra, an essential mathematical background for this book.
In conclusion this book is recommended both for initial acquaintance with statistical signal processing and its applications—and as a reference.
Copyright 1995- Analog Devices, Inc. All rights reserved.