EC4440 Statistical Digital Signal Processing

Modern methods of digital signal processing are developed in this course from a statistical point of view. Methods are developed for processing random signals through statistical data analysis and modeling. Topics include adaptive filtering, linear prediction, MA, and AR signal modeling, spectrum estimation and basic machine learning concepts. Techniques presented are applied to various engineering problems such as system identification, forecasting, and equalization. The algorithms introduced have direct applications in communication, sonar, radar systems signal processing, and modern Navy weapon systems. 

Prerequisite

EC3410 or EC3500 and MA3042 or consent of instructor

Lecture Hours

3

Lab Hours

2

Course Learning Outcomes

·       Develop the ability to characterize, analyze and extract information from random signals.

·       Learn to design FIR Wiener filters to extract information in the presence of additive noise.

·       Learn to apply basic adaptive filter structures such as the LMS and the RLS algorithm to extract signals from noisy environments. Understand the relative advantage and disadvantage of each adaptive method. Learn classic problems to which methods of adaptive filtering can be applied.

·       Learn how to apply the concept of mean squared filtering to spatial filtering.

·       Learn to apply adaptive filtering to extract the number and the direction of arrival of signals impinging on a linear array.

·       Learn classical methods of spectral estimation and their advantages and limitations to extract signal frequency information.

·       Learn basic principles of machine learning and their applications to engineering scenarios.

·       Understand the difference between supervised and unsupervised machine learning schemes.

·       Learn basic principles of neural networks, support vector machines, and k-means and clustering algorithms.