Scholarlink

<p>Structured Query Language injection (SQLi) attack detection is considered one of the most critical vulnerabilities in web applications, where an attacker injects malicious Structured Query Language (SQL) commands in the input fields. This causes an inconvenience to users and disrupts application functionality. To reduce errors and ensure the continued integrity and performance of web applications, a new method for SQLi attack detection is developed in this research. Here, Federated Learning (FL) is utilized to detect SQLi attacks to enable collaborative, decentralized model training across numerous devices without the requirement to share sensitive data. Initially, the submitted SQL queries are transformed into binary sequences. Next, one-hot encoding techniques, like Mutual Information, Entropy, Correlation coefficient, Information Gain, and Gini index, are applied to transform the query to a one-hot vector. Then, the processed data is fed to SQLi attack detection using Temporal Convolutional Long Short-Term Memory with Giant Magnificent Frigatebird Optimization (TempCLSTM_Gi-MFO). Moreover, TempCLSTM is trained by devising Giant Magnificent Frigatebird Optimization (Gi-MFO), which integrates Magnificent Frigatebird Optimization (MFO) and Giant Trevally Optimizer (GTO). Therefore, the TempCLSTM_Gi-MFO achieved an accuracy of 91.847%, F1-score of 91.680%, recall of 92.485%, and precision of 90.889% with a time step of 25&#xa0;s.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Structured query language injection attack detection via giant magnificent frigatebird optimization and federated learning

  • 期刊论文

  • 发表日期：

    2026年1月11日

查看全文
  • Madhavi Perla,
  • Valli Kumari Vatsavayi

摘要

Structured Query Language injection (SQLi) attack detection is considered one of the most critical vulnerabilities in web applications, where an attacker injects malicious Structured Query Language (SQL) commands in the input fields. This causes an inconvenience to users and disrupts application functionality. To reduce errors and ensure the continued integrity and performance of web applications, a new method for SQLi attack detection is developed in this research. Here, Federated Learning (FL) is utilized to detect SQLi attacks to enable collaborative, decentralized model training across numerous devices without the requirement to share sensitive data. Initially, the submitted SQL queries are transformed into binary sequences. Next, one-hot encoding techniques, like Mutual Information, Entropy, Correlation coefficient, Information Gain, and Gini index, are applied to transform the query to a one-hot vector. Then, the processed data is fed to SQLi attack detection using Temporal Convolutional Long Short-Term Memory with Giant Magnificent Frigatebird Optimization (TempCLSTM_Gi-MFO). Moreover, TempCLSTM is trained by devising Giant Magnificent Frigatebird Optimization (Gi-MFO), which integrates Magnificent Frigatebird Optimization (MFO) and Giant Trevally Optimizer (GTO). Therefore, the TempCLSTM_Gi-MFO achieved an accuracy of 91.847%, F1-score of 91.680%, recall of 92.485%, and precision of 90.889% with a time step of 25 s.