Abstract
In this paper we analyze the problem of multiple fault detection in a network of functions (cells implementing functions). It is shown that all multiple stuck-at-faults and all single bridging faults are detected by a test set which detects all multiple faults in every cell. In general it is shown that a number of test patterns is proportional to the number of lines in a network. For homogeneous trees the number of test patterns is shown to be linear with number of input variables. Upper bounds on number of test patterns is presented.