Programmed Search in Adaptive Systems

An investigation is conducted of the programming of search by discrete data systems over a space volume. The distribution of search effort which leads to the greatest attainable information rate on a contact, or probability of its retention by maximizing the probability of a positive interrogation within some designated time interval is determined. An estimate of p, the relative frequency of positive interrogations of an individual contact, is utilized as a basis for adjustment of the sampling rate in accordance with results of the search program optimization study. It is shown that in the case of Rayleigh signal sources, the rate of convergence to a stable estimate of p is greatest if a search program characterized by a rapid sampling rate and a consequent low probability of detection on the individual trial is employed. If, following a sequence of observations, the statistical distribution of the position of the contact may be ascertained with some confidence, a suitably restricted space volume may be used in the allocation of search. It is shown that, if an extremal exists, the uniformly interrogated search field, which is optimal in the sense that the product of the containment and detection probabilities is greatest, is defined by an equiprobability density of location contour.