The summation CFAR detector is widely used for the detection of narrowband signals. The normalized detection threshold determines the detection performance and must be appropriately chosen in practical applications. However, numerical problems often occur in the theoretical computation of the normalized detection threshold, particularly when channel power estimates are summed over a large number of overlapped input data blocks. This paper shows that the correlation between power estimates at the output of an FFT filter bank for successive input data blocks can be neglected for most common windows when the overlap ratio is less than or equal to 1/2. Under this constraint the normalized detection thresholds computed for overlapped and non-overlapped input data blocks are practically identical and the results for the probability of false alarm for a given threshold derived for non-overlapped input data blocks are applicable to overlapped input data blocks. This substantially simplifies the problem of computing the normalized detection threshold for overlapped input data blocks.

Additional Metadata
Keywords Constant false alarm rate, Detection and estimation, Fast fourier transform, Window
Conference 5th IASTED International Conference on Circuits, Signals, and Systems, CSS 2007
Citation
Wang, S. (Sichun), Inkol, R. (Robert), Rajan, S, & Patenaude, F. (François). (2007). Threshold computation for the summation CFAR detector: Non-overlapped versus overlapped FFT processing. In Proceedings of the Fifth IASTED International Conference on Circuits, Signals, and Systems, CSS 2007 (pp. 115–119).