基于信息论的程序可靠性分析

Abstract

This paper proposes a novel approach to program reliability analysis leveraging concepts and techniques from information theory. Traditional program reliability methods often rely on empirical testing and statistical modeling, which can be computationally expensive and may not fully capture the inherent uncertainty and error propagation within a program's execution. We introduce the concept of "program entropy" to quantify the uncertainty and error propagation risk associated with a program's logic. Furthermore, we explore the application of channel coding principles to enhance program robustness against interference and errors during execution. The core mechanism involves framing program execution as an information transmission process, allowing us to apply information-theoretic tools for rigorous analysis. This approach offers a theoretical framework for assessing and improving program reliability beyond the limitations of conventional methods. The key contribution lies in integrating information theory's insights into the traditionally deterministic field of software reliability.