Short-Period Noise Envelope Statistics: A Basis for Envelope Detector Design

Abstract

This report focuses on the use of certain detection statistics, in particular the instantaneous amplitude or envelope, and the instantaneous power, in the design of a controlled false alarm rate detector. To achieve false alarm rate control, the detection statistic must be stationary, but need not be Gaussian. Parameters of a Gaussian process can be conveniently transformed into a stationary, normalized detection statistic. For an envelope detector the normalization consists of dividing the envelope by a long-term (1 to 2 minutes) estimate of the RMS noise level; the instantaneous power is normalized by dividing it by the long-term average power. The stationarity of the normalized envelope is demonstrated with one hour of uninterrupted, Korean Seismic Research Station single-site, short-period noise, by compiling distributions of the base- ten logarithm of the normalized envelope. The envelopes closely follow the Rayleigh distribution which is the theoretical distribution for envelopes of a Gaussian process. Retaining this model at the 5% significance level establishes adequate control of the false alarm rate.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 26, 1978
Accession Number
ADA066713

Entities

People

  • Rudolf Unger

Organizations

  • Texas Instruments

Tags

DTIC Thesaurus Topics

  • Air Force
  • Algorithms
  • Data Science
  • Detection
  • Detectors
  • Distribution Functions
  • Frequency
  • Frequency Bands
  • Gaussian Processes
  • Goodness Of Fit Tests
  • Information Science
  • Probability Density Functions
  • Pulsed Power
  • Standards
  • Statistics
  • Surveillance
  • Waveforms

Readers

  • Approximation Theory.
  • Seismology