SEQUENTIAL SEARCH AND DETECTION
Abstract
The objective of this study was to obtain and evaluate strategies to be used in certain general search situations. These strategies minimize the expected cost of search and resulting decisions and are sequential in the sense that a decision at any time is dependent upon what has been observed up to that time. The first situation studied led to the formulation of a minimum expected cost sequential hypothesis test. The target is either present in the region of interest with a priori probability P , or not with probability 1 -P. Knowing the value of P , at fixed intervals of time the searcher must either make a terminal decision (i. e. decide that the target is present, or not present) or make a measurement of a random variable that has a probability density function which depends upon whether or not the target is present. A cost structure is given which assigns costs to wrong terminal decisions, as well as a cost (which depends upon whether or not the target is present) for the taking of a measurement. The second part of this study involves a situation in which it is assumed that the target arrives at some random time (the raid-recognition problem).
Document Details
- Document Type
- Technical Report
- Publication Date
- May 15, 1964
- Accession Number
- AD0601641
Entities
People
- Stephen M. Pollock
Organizations
- Massachusetts Institute of Technology