Track-before-detect performance for a high PRF search mode

The performance of a four-stage track-before-detect algorithm used for an X-band high pulse repetitive frequency (PRF) search mode is evaluated. A dynamic programming algorithm is used to noncoherently integrate over resolution cells lying along each potential target path, and path sums are compared to a threshold to determine if a target is present. Given a threshold setting that results in a false track generation rate of 10/sup 6/ per resolution cell, track detections are obtained with probability 0.5 at a signal-to-noise ratio of about 8.7 dB for a target closing at 2.5 nmi per revisit time, assuming a Swerling I or II Rayleigh signal plus noise model as measured after transforming to Doppler-range space. For the same threshold setting, 90% of the tracks were detected at signal-to-noise ratios (SNRs) of 9.6 dB or smaller. If the closing range is decreased to 0.25 nmi per revisit time, then track detections occur with probability 0.5 at a SNR of 8.1 dB, and 90% of the tracks were detected at SNRs of 5.5 dB smaller.<<ETX>>