Range resolution of targets

The problem of resolving targets in range is formulated as a hypothesis-testing problem. A generalized likelihood approach is developed and very good results are obtained, e.g., targets separated by more than a pulsewidth can almost always be resolved. Specifically, simulations showed that when using a sampling rate of 1.5 samples per pulsewidth, two 20 dB nonfluctuating targets can be resolved at a resolution probability of 0.9 and a false-alarm probability of 0.01 at separations varying between 1/4 and 3/4 of a pulsewidth, depending on the relative phase difference between the targets. The effect of increasing the sampling rate was investigated and it appears that there is only marginal benefit in increasing the sampling rate beyond 1.5 samples per pulsewidth. Finally, the generalized likelihood approach was compared to several easily implemented adhoc approaches and an adhoc approach involving fitting a pulse shape to the data is only slightly (approximately 10 percent) less accurate than the likelihood approach.