Secure Resource Allocation Optimization in Cloud Computing Using Deep Reinforcement Learning

Abstract

This paper presents a new security resource allocation optimization framework for a cloud computing environment using deep learning (DRL). The framework addresses the key issues of measuring resource efficiency with security policies in a cloud environment. A security-aware DRL model is developed, incorporating a comprehensive reward function that integrates resource optimization objectives with security constraints. The proposed architecture implements a multi-layer neural network specifically designed for processing complex cloud system states and security metrics. The framework features an adaptive security system that continuously evaluates and responds to potential threats while maintaining efficient resource allocation patterns. The experimental results show a significant improvement over traditional methods, achieving a 17.6% increase in resource utilization and maintaining 95% security quality. The system exhibits robust performance under various attack scenarios, with threat detection rates exceeding 94% across different security breach attempts. Performance evaluations conducted on a large-scale cloud platform validate the framework's effectiveness in real-world environments, showing a 45.3% reduction in response time compared to baseline methods. The plan balances the trade-off between resource optimization and security monitoring, providing practical solutions for secure cloud distribution in today's devices.