Development of Reinforcement Learning and Pattern Matching (RLPM) Based Firewall for Secured Cloud Infrastructure

Cloud computing infrastructure is typically intended to store and deliver sensitive data and high performance computing resources through the internet. As the utility of cloud computing has increased to larger extend because of its sophisticated services, the security breaches also growing proportionately in terms of third party attacks. In order to mitigate the modern security attacks in the cloud environment, the traditional firewall rules and packet filtering methods are absolutely insufficient. Hence, a Deep Packet Inspection based firewall (RLPM) is developed to block the malicious attacks by validating the payload signature of arriving packets. RLPM combines the potential of Reinforcement Learning (RL) and parallel fast pattern matching simultaneously and it converges to an optimal solution at the earliest. RL method efficiently learns the environment and process the payload signature in a parallel manner. A two-way pattern matching algorithm is integrated with RL approach that validates the signature towards attaining the quick decisions. The performance results show that the RLPM is better as compared to the existing methods in terms of Response time, throughput and malicious attack blocking. As the firewall is deployed and tested in a real cloud computing environment, the response time is found to be 10% lesser while throughput is also increased about 10% than the existing state-of-the-art-methods.