Kalman filtering with real-time applications (Q1098815): Difference between revisions

This book is another addition to the multitude of books/monographs recently published in this field. It aims at ``disclosing the mystery (of discrete Kalman-filtering) by presenting a fairly thorough discussion of its mathematical theory and applications to various elementary real-time problems,'' because according to the authors ``the mathematical theory of Kalman filtering and its implications are not well-understood even among many applied mathematicians and engineers.'' After such questionable claims, the text starts with a review of some results from matrix and probability theories and least squares optimization. Chapter 2 involves the derivation of the unbiased, mean square optimal, one-step state predictor (Kalman filter) for a linear discrete-time system with a linear measurement equation using the least squares optimization results introduced in chapter 1. In chapter 3, the Kalman filter is rederived, this time using orthogonal projection methods. Chapter 4, involves derivation of the filter for correlated input and output noises, whereas in chapter 5, a technique for treating, the colored noises is given. Chapter 6 is on limiting properties of Kalman filter. Based on controllability and observability assumptions (which are not relaxed anywhere in the book), existence of the unique, positive definite steady state solution (which is independent of the initial condition) of the Riccati equation is proved. Also geometric rate of convergence to this solution is shown. In chapter 7, two algorithms are given for finding the Kalman gain, and then these are combined into one algorithm. Chapter 8 contains the extended Kalman filter and its application to satellite orbit estimation and adaptive system identification. In chapter 9, a technique is presented to decompose the steady state filter into subsystems for possible computational simplicity and order redution. In the last chapter, important topics such as smoothing, identification, continuous-time filtering, stochastic control, etc. are very lightly touched upon. Although this book may not be suitable for adoption as a graduate textbook because of its limited coverage of the estimation and detection field and its dependence on lifeless academic examples for illustration of the theory, it may, however, be of some use to the practicing engineer, who is aware of the potential applications of this theory and needs to brush up his/her knowledge in this field, as a self-study guide.