www.radartutorial.eu www.radartutorial.eu Radar Basics

Correlation

Correlation is a method for comparing two signal waveforms. It describes how closely the signals match at a particular time shift. In signal processing, this method is mainly used to detect known signal patterns, determine differences in propagation time, and identify periodic signals hidden in noise. Correlation is therefore an important analysis method in the time domain.

In radar signal processing, a received signal is often compared with a known reference signal. To do this, the two waveforms are shifted relative to each other, either conceptually or by calculation. The correlation function reaches a maximum at the time shift where the signals match most closely. This maximum indicates the time delay between the two signals. For continuous signals, the comparison is performed by multiplying the corresponding signal values and then integrating the products over the time interval under consideration:

Equation (1) (1)

For discrete or digital signals, the corresponding signal values are multiplied and the resulting products are then added together:

Equation (2) (2)

Here, τ or m, respectively, represents the time shift between the two signals. The closer the signals match at a particular time shift, the more pronounced the corresponding correlation peak becomes.

When a signal is compared with a time-shifted copy of itself, this is called autocorrelation. It can be used, for example, to detect recurring signal components or periodic patterns. Comparing two different signals, on the other hand, is called cross-correlation. In the ideal case, a radar echo is a delayed and attenuated copy of the transmitted signal. The comparison between the actual received signal and the transmitted signal therefore uses cross-correlation, which, in this ideal case, has the shape of a shifted and scaled autocorrelation of the transmitted signal.

The reference signal is usually known, for example as a stored copy of the transmitted radar signal. It may also be normalized to a particular amplitude or energy for the correlation calculation. However, an ordinary correlation function is not limited to a maximum value of 1. Only with appropriately normalized correlation can the degree of agreement be evaluated using a defined range of values.

Practical applications of correlation include the matched filter and range measurement using noise radar. Since the echo arrives later than the transmitted signal, the signal propagation time can be determined from the position of the correlation peak. This propagation time can then be used to calculate the range to the reflecting radar target.

Source: