Model-based signal processing

This course offers an introduction to estimation and detection theory with emphasis on applications in signal processing and commmunications. It provides an overview of the main tools to extract desired information from measured signals under a statistical signal processing framework. Fundamental limitations on the estimation performance is discussed. A brief introduction to detection theory and classification is presented.

Content

Minimum Variance Unbiased Estimation and the Cramer-Rao Lower Bound, Best Linear Unbiased Estimators, Maximum Likelihood Estimation, Least Squares, Method of Moments, Bayesian Estimation, Wiener and Kalman Filters, Introduction to Detection Theory, Likelihood Ratio Tests, Model Order Selection.

Learning Outcomes

• Understand how to model experimental data in simple situations and make suitable assumptions on the random components in the data
• Understand the fundamental limitations of statistical data models and how these can be used to design experiments
• Understand the trade-offs involved in parameter estimation
• Know how to choose a suitable estimation approach for a given data model and derive the resulting estimator
• Know how to implement a variety of estimators on a computer
• Know how to evaluate the performance of estimators and compare different approaches
• Have a brief knowledge about detection techniques and selection of a suitable model order

Prerequisite

Basic knowledge of probability theory and stochastic processes.