1. Introduction
We use the term computer system to describe collections of
computational resources. These resources may communicate with
each other or may comprise independent computers that interact
only with human programmers and operators. With increasing
frequency, computer systems are relied upon in applications where
the successful completion of tasks and activities is especially
critical. In many cases, a computer system is required to func-
tion without error or within certain efficiency bounds. The
extent to which the systems fulfills such requirements is used to
define the re1iability or performance of the system. In highly
reliable or efficient systems, system behavior is measured against
its response to external input which lies within a certain range
that has been specified as acceptable. Input lying outside the
specified range may cause the system to fail or to become
inefficient, but, since the system meets its requirements on the
input for which it was designed, its reliability or level of per-
formance are not affected by such input.
There are applications, however, for which such notions as
reliability are unsuitable. In these applications, the range of
specified input is determined by an adversary or enemy whose goals
is to subvert the successful functioning of the system. Thus,
1
http://dx.doi.org/10.1090/psapm/029/9880
Previous Page Next Page