A Rule-based Approach For Simulating Errors In Discrete Sequential Processes

Discrete sequential processes consist of sequences of actions or symbols, each of which is often associated with a relative frequency of occurrence. Since correct and incorrect sequences can be identified by formal bottom-up parsing techniques, a top-down approach can be used in conjunction with a stochastic grammar to generate sequences in a simulation environment. Such an approach combines recognition and simulation of sequences and is particularly useful when test patterns are needed for a complex recognition system. This paper presents a Prolog-based simulation system wherein the productions of a stochastic grammar are expressed in an English-like IF-THEN format so they can easily be edited to simulate parsing errors leading to incorrect sequences. Applications involving simulated errors in communication systems are discussed.