Multirate statistical signal processing (Q871288)

The book presents a set of results, obtained by a thorough study of a variety of signal processing problems that mainly involved processing low-rate components of a full-rate information bearing signal. The author is developing mathematical models for these problems, from a statistical viewpoint, where the original full-rate signal is assumed to be a stationary random process. This approach has allowed finding unique and stable solutions to the ill-posed inverse problem of reconstructing the samples of the original signal, obtained from the available low-rate observations. It also has allowed posing and solving other interesting problems, such as estimating the spectrum of a full-rate signal from low-rate observations, quantifying the information content of individual low-rate measurements and specifying multirate filter banks that lead to scalable decompositions. The developed statistical theory presented in the book is very general in the sense that it can be used to pose and solve multirate signal processing problems, involving both FIR and IIR filter banks. Furthermore, there are no restrictions on the number of channels and/or the down-sampling ratio in each channel. These are important advantages that distinguish the statistical approach from conventional deterministic formulations considered in the past. The main goal of the author in writing this book is to inspire the reader to initiate his own research and add to the theory of multirate statistical signal processing. Having this goal in mind, the author presents background material, key principles, potential applications and open research problems while striking the appropriate balance between clarity and brevity. Having this in mind, the author created the content of the book in nine chapters, as follows: 1. Introduction, where digital multirates signal processing, its applications and multirate statistical signal processing is introduced. 2. Backgrounds, which reveals the essence and content of inverse and ill-posed problems, inequality and information measuring, statistical inference and stochastic processes. 3. Multirate Spectrum Estimation is a chapter, where are presented the mathematical modeling of the problem, the Maximum Entropy principle, a geometric interpretation, properties of the Maximum Entropy solution, computing the Maximum Entropy solution, simulated examples and open problems. 4. Multirate Signal Estimation, a chapter which is devoted to the stochastic least-square estimation problem and its solution. Linear least-squares estimation and estimator matrix computing are particularly considered. Some simulated examples are given and multirate least-squares estimation is considered in practice. 5. Multirate Time-Delay Estimation is the focus of the next chapter. Time-delay estimation techniques in multirate system and multirate sensors that allow time-delay estimation are under consideration. Laboratory experiments are depicted and multirate sensor fusion in the presence of time-delay is presented. 6. Optimal Multirate Decomposition of Signals is the next chapter. A review of FIR filter banks is given. The scalability and optimality are under consideration. An illustrative design example, in this sense, is provided. A comparison between optimality and Subband Coding optimality is considered next. Some open problems are stated. 7. Information in Multirate Systems is a chapter, where the Information is interpreted as a distance from uniform spectrum. An illustrative example is provided. The topics of redundancy and scalability in terms of information are considered. 8. Distributed Algorithms is the final theme of the book. The need for distributed algorithms is reasoned. Spectrum estimation as a convex feasibility problem is presented and a solution using generalized projections is given. Distributed algorithms based on local generalized projections are depicted. Some open problems are described. 9. An epilogue and rich list of references are provided at the end. The logical sequence of the chapters provide a systematic introduction to the theory of statistical multirate signal processing. As this theory is still far from complete, the reader has the opportunity to examine the initial results presented in the book and extend them in to the directions in which this book points. The book will be of interest for all specialists working in the area of digital signal processing. It will be very useful for the students, postgraduates and researchers, which are looking for new directions of their studies.