关键词: 云计算, 深度学习, 协议, 监测系统, 恶意代码

Abstract: The development of various types of web applications has enabled the application range of the HTTP protocol to continue to expand. Due to the flexibility and diversity of Web technologies, new attack methods for Web applications are constantly evolving and evolving. Malicious behaviors such as XSS cross-site scripting attacks and database injection attacks are increasingly reflected in HTTP requests. The traditional security defense system responds to new challenges all the time.In order to cope with the new changes of Web security, this paper proposes a method to establish the HTTP protocol security monitoring model, the classification algorithm training for the model, and the detection of HTTP access data categories. Combining the characteristics of malicious behavior in HTTP request, the change of malicious feature path, and the difficulty of monitoring, explore the use of cloud computer technology to analyze HTTP request format and malicious features, automatically generate sensitive word database, and adopt feature entropy-based feature selection algorithm. In combination with deep learning technology, the classification algorithm is trained to train the security detection model, and then a monitoring system for improving the security of the HTTP protocol is proposed.

Key words: cloud computing, deep learning, protocol, monitoring system, malicious code