An operator perspective on signals and systems (Q1039400)

This book covers a number of fundamentals of signal and system theory mostly concentrating on state space methods combined with application of operator techniques. Based largely on state space methods, this text utilizes fundamental linear algebra and operator techniques in order to solve spectral estimation, factorization and interpolation problems in control design and signal processing. The spectral factorization problems in control theory can be studied in terms of Riccati equations arising in linear quadratic control theory and Kalman filter theory. In this monograph the authors employ operator theory to prove the existence of a solution to some factorization and spectral estimation problems in a very general setting. In particular, the spectral factorization problem is addressed in terms of finite sections of Toeplitz matrices and in terms of Riccati factorization. It is to be noted that the whole content of this book is well structured in five parts and it well balances the different aspects of the theory. After an introduction to the classical results from basic operator theory such as Wold decomposition, Toeplitz and Laurent operators and Naimark representation theorem presented in the first part, the authors move to the solution of spectral factorization problems in square and non-square cases with particular application to the sinusoid estimation problem in signal processing. Isometric representation and factorizations (Cholesky and inner-outer algorithms) are also discussed in the second part. The third part is devoted to the fundamentals of the broadly used Kalman and Wiener filtering techniques based on algebraic Riccati equations and factorization. One finds results on interpolation theory in part four, where both tangential and contractive Nevanlinna-Pick interpolations are discussed. An appendix in the final part reviews the classical state space methods and terminology used throughout this monograph. The bibliography has 204 entries including a small index. This monograph can be considered as the second volume of the monograph [\textit{M. J. Corless} and \textit{A. E. Frazho}, Linear systems and control. An operator perspective. Pure and Applied Mathematics, Marcel Dekker 254. New York, NY: Marcel Dekker (2003; Zbl 1050.93001)], where the authors restrict themselves to the finite-dimensional situation not handling the infinite-dimensional cases as in this book. The missing link here is how the Kalman filter can be obtained from input-output data using linear algebra tools. Once these states are known the identification problem becomes a linear least squares problem in the unknown system matrices. The solution for this can be formulated in terms of subspace identification algorithms which conditionally linearize the problem and identify input-output models. For more details on subspace identification theory readers may refer to the monograph by [\textit{P. Van Overschee} and \textit{B. De Moor}, Subspace identification for linear systems. Theory, implementation, applications. Dordrecht: Kluwer Academic Publishers (1996; Zbl 0888.93001)]. A number of examples using \texttt{Matlab} are also given. It is of interest for studying and teaching mathematical fundamentals of engineering solutions developed for signals and systems analysis, and can be recommended for both engineers interested in deeper understanding of modern and classical mathematical results behind such conventional tools as the Kalman filter, and pure mathematicians for taking a wider view on solutions recently deployed for signals and systems.