Scholarlink

<p>With the rapid development of the Internet of Things and smart devices, traditional security detection methods have been unable to cope with emerging security issues in embedded systems. This paper combines convolutional neural network (CNN) with deep reinforcement learning (DRL) to apply a model based on convolutional reinforcement learning (CRL) model. This paper first provides a security monitoring basis through multi-dimensional modeling analysis, and then uses CNNs to extract system operation data features. The DRL mechanism dynamically adjusts the defense strategy to enhance the system’s responsiveness and adaptability to complex attacks. Finally, an adaptive defense mechanism is designed to improve the system’s security and stability. Experimental results show that under severe attack intensity, the precision, recall, and F1 value of the CRL model proposed in this paper reach 90%, 89–89.5%, respectively, which are significantly better than the existing DRL, CNN, LSTM (Long Short-Term Memory), and SVM (Support Vector Machine) models. This fully demonstrates the excellent performance of the CRL model in terms of performance and stability.</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

Embedded computer software security monitoring system based on dynamic intelligent algorithm

  • 期刊论文

  • 发表日期：

    2025年10月3日

查看全文
  • Weiqiang Hu

摘要

With the rapid development of the Internet of Things and smart devices, traditional security detection methods have been unable to cope with emerging security issues in embedded systems. This paper combines convolutional neural network (CNN) with deep reinforcement learning (DRL) to apply a model based on convolutional reinforcement learning (CRL) model. This paper first provides a security monitoring basis through multi-dimensional modeling analysis, and then uses CNNs to extract system operation data features. The DRL mechanism dynamically adjusts the defense strategy to enhance the system’s responsiveness and adaptability to complex attacks. Finally, an adaptive defense mechanism is designed to improve the system’s security and stability. Experimental results show that under severe attack intensity, the precision, recall, and F1 value of the CRL model proposed in this paper reach 90%, 89–89.5%, respectively, which are significantly better than the existing DRL, CNN, LSTM (Long Short-Term Memory), and SVM (Support Vector Machine) models. This fully demonstrates the excellent performance of the CRL model in terms of performance and stability.