Data Flow Anomaly Detection

Abstract
The occurrence of a data flow anomaly is often an indication of the existence of a programming error. The detection of such anomalies can be used for detecting errors and to upgrade software quality. This paper introduces a new, efficient algorithm capable of detecting anomalous data flow patterns in a program represented by a graph. The algorithm based on static analysis scans the paths entering and leaving each node of the graph to reveal anomalous data action combinations. An algorithm implementing this type of approach was proposed by Fosdick and Osterweil [2]. Our approach presents a general framework which not only fillls a gap in the previous algorithm, but also provides time and space improvements.

This publication has 4 references indexed in Scilit: