AbstractsMathematics

Nonparametric sequential detection

by James Charles Fowler




Institution: Missouri University of Science and Technology
Department:
Year: 1968
Record ID: 1532107
Full text PDF: http://hdl.handle.net/10355/26415


Abstract

"This dissertation extends the theory of the Wilcoxon-Mann-Whitney U statistic so that this statistic can be used to perform sequential tests on hypotheses. This sequential test procedure makes use of a sequential ranking procedure similar to the one first introduced by Parent. The operating-characteristic function and average number of samples function for this new test are calculated as a function of the signal to noise ratio. The test is then shown to be efficient for several forms of alternatives with an efficiency of 95% against the Wald Sequential Probability Ratio Test for a constant signal in normal noise. Finally, the test procedure is modified so that it is capable of making measurements on the channel in order to adapt itself to changes in the channel characteristics. Simulation results are presented to show that this adaptive detector can operate with low probability of error" – Abstract, p. i.