Theory of the FFT Filter Bank-Based Majority and Median CFAR Detectors


  1. Wang, S.
  2. Inkol, R.
  3. Rajan, S.
  4. Patenaude, F.
Corporate Authors
Defence R&D Canada - Ottawa, Ottawa ONT (CAN);Communications Research Centre, Ottawa ONT (CAN)
The FFT filter bank-based majority and median constant false alarm rate (CFAR) detectors are very useful alternatives to the well-known FFT filter bank-based summation CFAR detector. However, for overlapped input data blocks, the theoretical performance analysis of the majority and median CFAR detectors is considerably more difficult than that of the summation CFAR detector, the main reason being that formulas for computing the probability of false alarm for a given threshold can not be easily derived for them. This technical report presents new results on threshold computation for the FFT filter bank-based majority and median CFAR detectors for both overlapped and non-overlapped input data blocks. The most important contribution of this report is the successful computation of the joint probability density function of the output power levels of the FFT filter bank for overlapped white Gaussian noise input blocks. As a corollary, it is shown that the statistical correlation of the output power levels of the FFT filter bank can be neglected for several commonly used windows, such as the Blackman and Blackman-Harris windows. Complete mathematical derivations are provided for all the major theorems presented in the report.

Il y a un résumé en français ici.

Report Number
DRDC-OTTAWA-TR-2007-088 — Technical Report
Date of publication
01 May 2007
Number of Pages

Permanent link

Document 1 of 1

Date modified: