Reinforcement Learning and Anomaly Detection as a Defense for Edge-Based DDoS Attacks

Edge computing has emerged as a solution to minimize cloud computing’s latency by placing servers physically closer to the users. However, unlike centralized servers utilized in cloud computing, edge servers are limited by resources and as a decentralized network, is vulnerable to cyberattacks such as Distributed Denial of Services (DDoS) attack. This paper proposes a model based on AWS’s architecture with additional components. The added components are reinforcement learning (RL) based resource management for optimizing VM instances and granting defence systems with more resource, the implementation of Akamai Prolexic alongside with an AI-driven threat detection and response system to mitigation malicious traffic before it reaches internal system. The proposed model is finically costly and difficult to implement for small-medium enterprises however, in large enterprises like AWS and Google Cloud, the performance gains several advantages such as adaptability, scalability, a proactive defence system, etc.