Accuracy relates to a measurement or prediction being close to the true value of target parameters. Precision relates to the fineness of the measurements, which may not be very accurate, but could be quite precise.

Target parameters for which accuracy is important include range, angle, Doppler, and amplitude. Accuracy varies as a function of range. At long range, thermal noise effects tend to dominate.

At intermediate ranges, accuracy is dominated by the instrumentation errors (relatively constant vs. range). At short ranges, angle glint effects can dominate since the angular extent of the target increases inversely with range.

The accuracy of a given measurement due to thermal noise is given by σ = K/√SNR where K has the same dimensions as the measurement, but is also inversely proportional to the effective width in the other domain of a Fourier transform (FT) pair (i.e., range or time has frequency or bandwidth as its FT pair).

Hence, the K for a range measurement is inversely proportional to bandwidth and the K for a Doppler frequency measurement is inversely proportional to time extent of the waveform, that is, takes a long time to discern small differences in frequency.

Since an antenna pattern is the FT of its aperture distribution, the K for angle accuracy is inversely proportional to effective aperture width. Resolution pertains to the question: Is there one target present or many? If two targets are resolved in range (i.e., well separated compared to the compressed pulse width), there will be two distinct returns.

As the targets get closer together, the returns begin to merge such that it is difficult to tell if there is one or two since the thermal noise tends to distort the combination. The presence of a dip between them yielding two peaks will depend on the relative phases of the two pulses.

Typical resolution algorithms include the classical inflection or dip approach, as well as template matching algorithms that look for differences compared to the known response of a single point target. Multipath and thermal noise will affect the probability of correctly resolving two targets in range when two targets are present as well as the probability of false splits (i.e., claiming that two targets are present when only one is actually present).

Similar algorithms are used for resolution in angle when the beam scans past the target. If frequency diversity is used on different prfs, this will cause the amplitude to fluctuate as the beam scans past the target making it even more difficult to determine whether there is one or two targets present.

With a monopulse radar, one can examine the imaginary part of the complex mono pulse ratio to determine if more than one target is present. A single target creates a quadrature value of zero, and multiple targets can create a nonzero value.

No comments:

Post a Comment

Related Posts Plugin for WordPress, Blogger...