EC4570 Signal Detection and Estimation

Principles of optimal signal processing techniques for detecting signals in noise are considered. Topics include maximum likelihood, Bayes risk, Neyman-Pearson and min-max criteria and calculations of their associated error probabilities (ROC curves). Principles of maximum likelihood, Bayes cost, minimum mean-square error (MMSE), and maximum a posteriori estimators are introduced. Integral equations and the Karhunen-Loeve expansion are introduced. The estimator-correlator structure is derived. Emphasis is on dual development of continuous time and discrete time approaches, the latter being most suitable for digital signal processing implementations. This course provides students the necessary foundation to undertake research in military radar and sonar systems. 

Prerequisite

EC3410 or EC3500

Lecture Hours

4

Lab Hours

0

Course Learning Outcomes

·       Given the problem parameters, the student will be able to define the strategy to be used and designs the appropriate detector and determine the correct threshold to be used with the detector.

·       Given the detector and threshold, the student will be able to obtain the statistics of the detection process, probability of detection, probability of false alarm, probability of miss, and probability of correct dismissal.

·       The student will be able to implement a non-parametric decision test.

·       The student will be able to implement a sequential hypothesis decision test, given data in the MATLAB environment.

·       The student will be able to derive a decision test to detect a discrete dynamic signal in noise.

·       The student will be able to assess the performance of simple estimation schemes using the ROC.