Current trends and research gaps in SQL injection simulation and mitigation for IaaS: A systematic review

Abstract

The rapid migration of enterprise operations to cloud environments has significantly escalated the risk of SQL injection (SQLi) attacks, necessitating more robust defence mechanisms for web applications. This study aims to provide a comprehensive evaluation of the current SQLi landscape, focusing on attack typologies, detection methodologies, and mitigation efficacy. Using the PRISMA framework, a systematic literature review was conducted on 38 peer-reviewed studies published between 2021 and 2025, indexed in the Scopus and Web of Science databases. The analysis reveals that 60% of research focuses on general SQLi threats, while specialized subtypes remain under-investigated, with 29% addressing blind/ordered injections and only 3% targeting specific in-band or out-of-band vulnerabilities. Regarding detection, the field is dominated by static analysis (32%), followed by deep learning (21%) and hybrid approaches (21%), whereas dynamic and rule-based methods comprise only 10% and 3%, respectively. For mitigation, deep learning (48%) and machine learning (37%) emerge as the primary defences, with reinforcement learning and algorithmic solutions each contributing 5%. The results demonstrate that while current models excel in simulated environments, there is a critical gap in validating traffic within live Infrastructure-as-a-Service (IaaS) platforms. Furthermore, the findings highlight a lack of resilience against adversarial attacks and imbalanced datasets. This study concludes that future research must pivot toward adaptive, real-time detection frameworks and more realistic cloud simulation models. These insights are vital for developers and security architects aiming to safeguard cloud-native applications against evolving injection techniques, ultimately providing a roadmap for more resilient cybersecurity infrastructures.