Triggered Correlation

This paper shows to what extent an average-response computer can be utilized for computing a cross-correlation function. This type of computer needs synchronization pulses, and the simplest methods of computation are those in which these pulses are directly derived from one of the signals (triggered correlation). The first method is to generate a synchronization pulse whenever the signal crosses a pre-set threshold in any direction. In this case, the computer output function is shown to be proportional to the true correlation function, for Gaussian signals. In a second method, synchronization pulses are produced when the signal crosses the threshold in a specified (e.g., positive) direction. Then the computer output is found to be contaminated by a systematic error, which, in turn, depends on the derivative of the correlation function. These two methods are described in detail, both with respect to the results and to the accuracy obtainable. Several other, less important, methods are only briefly described.