The purpose of this paper is to discuss an epoch detection procedure which is very useful for the resolution and detection of signals overlapping in time. An epoch is the beginning instant of a signal. The epoch detection procedure is based on the following hypotheses: On the null hypothesis H0that a certain instant t is not an epoch, analytical continuation exists at t, and one may predict the signal in the future based on past experience or vice versa. On the hypothesis H1that t is an epoch, the analytic continuation is disrupted at t. Based on this idea and the assumption of a Gaussian noise, a test statistic is derived from the maximum likelihood principle. The test statistic may be obtained at the output terminal of a linear filter. The performance of such a system is considered. Also discussed briefly are the cases of overlapping stochastic signals and overlapping radar signals. Some experimental results obtained from a digital computer are shown.
No takes yet. Share an insight, caveat, or question.
Tzay Y. Young (1965) studied this question.
Synapse has enriched 5 related papers. Consider them for comparative context: