Information and Inference for Point Processes

Estimation, Decision, and Adaptive Sensing
Erscheinungsjahr 2027
ISBN: 978-3-032-47214-4
Verlag: Springer

Buch, Englisch, Format (B × H): 155 mm x 235 mm

This book develops principled methods for sensor management when an uncertain scene is represented by a point process. This book shows how information theory, estimation bounds, population error measures, and decision theory can be used to compare and select sensing actions. It develops entropy and mutual information for point-process states, information contraction under observation and compression, Fisher information and Cramér-Rao and Van Trees bounds, dependence-aware precision, quadratic and Mahalanobis population errors, Bayes risk, and closed-loop and distributed sensing. Information measures quantify uncertainty reduction; estimation bounds assess attainable precision; population errors assess estimated configurations; and decision risk measures task consequences. The book explains when these criteria agree, when they differ, and how they guide multi-target surveillance, sensor scheduling, waveform and aperture selection, distributed fusion, point-source sensing, and communication-constrained networks. It is intended for researchers and doctoral readers in statistical signal processing, communications, tracking, sensor fusion, and adaptive sensing.

Zielgruppe

Professional/practitioner

Autoren/Hrsg.

Clark, Daniel

Fachgebiete

• Mathematik | Informatik EDV | Informatik Informatik Mathematik für Informatiker
• Technische Wissenschaften Elektronik | Nachrichtentechnik Nachrichten- und Kommunikationstechnik Signalverarbeitung
• Technische Wissenschaften Elektronik | Nachrichtentechnik Elektronik
• Mathematik | Informatik Mathematik Stochastik

Inhaltsverzeichnis

Introduction.- Part I: Statistical Populations, Observation, and Information.- Information, Search, and Sensor Management.- Point Processes as Statistical Populations.- Entropy and Relative Information.- Mutual Information in Sensor Networks.- Information Contraction Under Observation.- Part II: Scores, Precision, and Information Geometry.- Scores and Bartlett Identities.- Fisher Information and Cramér-Rao Bounds.- Bayesian Information and Van Trees Bounds.- Dependence, Covariance, and Information Geometry.- Part III: Error Geometry and Performance.- Quadratic Error for Random Measures.- Mahalanobis Geometry for Random Populations.- Performance Assessment for Multi-Target Systems.- Part IV: Decision and Adaptive Sensing.- Risk and Decision for Random Populations.- Information-Driven Sensor Management.- Closed-Loop and Distributed Sensing.- Conclusion.

Über Autor(innen)

Daniel E. Clark is the chair in Electronics and Computer Science at the University of Southampton. His research spans statistical signal processing, point processes, random measures, multi-object inference, sensing, estimation, and information theory. His original contributions include local entropy statistics for point processes, multi-sensor network information for multi-target systems, a Cramer-Rao bound for point processes, and functional Bartlett identities for spatial point-process likelihoods. His current work develops quadratic and precision-weighted errors for random counting measures and connects these methods to decision and adaptive sensing. He is the author of Random Point Sources: A Mathematical Theory of Point-Source Recovery, forthcoming in Springer’s Signals and Communication Technology series, and Field Theory for Point Processes, forthcoming in Birkhauser’s Progress in Probability.