Paper Info
Title
On the Shapiro-Wilk Test for the Detection of Pulsed Sinusoidal Radio Frequency Interference
Abstract
In this presentation, the Shapiro-Wilk test for normality (3) is analyzed, and its detection performance compared with that of the kurtosis test. The study considers only pulsed and continuous sinusoidal RFI as in (2); other RFI types may produce distinct conclusions regarding the relationship between the tests. The Shapiro-Wilk test involves a ratio of two distinct estimates of the sample variance, one of which is appropriate only if the data arises from a Gaussian distribution. Computation of the Shapiro-Wilk test statistic requires that a sort operation be performed on an observed data sample; such operations are feasible for implementation in digital hardware. Data discretization effects also play an important role in test performance for digital radiometer systems. Due to diføculties in predicting statistical properties of the Shapiro-Wilk test statistic, particularly for discretized data, Monte Carlo methods are used to estimate the probability of detection and false alarm as a function of RFI parameters. Two sample results are shown in the next Section.
Year
DOI
Venue
2008
10.1109/IGARSS.2008.4778951
IGARSS
Keywords
Field
DocType
noise,testing,statistical distributions,hardware,quantization,remote sensing,gaussian distribution,data mining,statistical analysis,shapiro wilk test,probability of detection,monte carlo method,monte carlo methods,radio frequency interference,monte carlo simulation
Normality,Shapiro–Wilk test,Computer science,Electromagnetic interference,Electronic engineering,Artificial intelligence,Microwave radiometry,Kurtosis,Computer vision,Monte Carlo method,Statistics,Quantization (signal processing),Statistical analysis
Conference
Citations 
PageRank 
References 
1
0.39
9
Authors
3
Name
Order
Citations
PageRank
Baris Guner1162.65
Mark T. Frankford2202.86
Joel T. Johnson3926116.14