Self-Healing Autonomous Software Code Development
Sandeep Kumar Jangam · International Journal of Emerging Trends in Computer Science and Information Technology · 2022
The complexity of modern software systems and their size have augmented the need for resilient, even adaptive and autonomous maintenance functionality. Manual error repair methods of code injection and repair, and traditional debugging procedures are imperfect in the context of fast-moving clouds like cloud-native applications, growth in edge computing, and autonomous systems, because they are usually slow and subject to inaccuracy. The problem of this research is the pressing need to develop self-healing software to do the detection, diagnosis and repair of faults in its own codebase without the involvement of a human being while it is executing. With the new possibilities in the fields of artificial intelligence, machine learning and program synthesis, we present a new model that can be used to perpetually track the behavior of the code and anomalies with the help of statistical and behavioral signatures and automatically achieve a solution by applying learned fixes. We represent a combination of deep learning-based fault localization, reinforcement learning with policy optimization and semantics-based code mutation in order to restore self-repair in real time. It has been tested on an assortment of open-source programs with shared pieces of software bugs, with a 78 percent success ratio of self-creating fixes and an average of a 32% decrease in mean time to recovery (MTTR) over all present automated ironing methods. The findings are used to show how the concept of incorporating autonomous healing properties into software systems has viability and efficiency in minimizing downtime, maintenance overloads and enhancing the reliability of software. This research establishes the basis of the development of the next generation of intelligent software whose behavior is not simply reactive, but also self-improving