Reinforcement Learning-Based Load Balancing with Large Language Models and Edge Intelligence for Dynamic Cloud Environments

Abstract

Traditional load balancers in cloud environments face significant challenges in managing traffic spikes, leading to increased latency and potential security vulnerabilities. This paper proposes a novel approach to cloud load balancing by integrating reinforcement learning, large language models (LLMs), and edge intelligence. Edge computing enables distributed decision-making, improving latency performance and user experience through localized data processing. AI-driven anomaly detection enhances security by continuously monitoring traffic behavior to identify and mitigate threats, while auto-scaling capabilities ensure scalability by adjusting server capacity in response to workload fluctuations. Our approach demonstrates significant improvements in throughput efficiency, security, and latency management compared to default configurations of AWS ELB, Azure Load Balancer, and GCLB. Despite these advancements, challenges such as dependency on proprietary cloud APIs and the need for improved multi-cloud interoperability remain. Future research should focus on enhancing AI/LLM adaptability, exploring advanced reinforcement learning techniques, and addressing security challenges through predictive analytics. This framework offers a robust solution to enhance performance, security, scalability, and operational efficiency in modern cloud-based applications.