SDN Based Adaptive Zero Trust Security for East-West Traffic

Abstract

Software defined networking (SDN) introduces centralized network control which separates the data plane, control plane and application plane by enabling programmability, scalability, and flexibility. Observing the existing system which mainly focuses on protecting external incoming and outgoing traffic between hosts by using traditional firewalls. New security risks are majorly associated with internal traffic, such as east west traffic. Majority of the new security threats have arisen in internal communication, such as east-west traffic. The project follows the NEVER TRUST ALWAYS VERIFY principle, which requires that each device's requests to be examined before granting access. The proposed system provides adaptive zero-trust security for the SDN environment to continuously validate all communication requests before granting access to the entire network in order to prevent default trust on each device, user, or system. Because the proposed hybrid machine learning management model combines supervised and unsupervised algorithms to prevent the attacks. The XG-Boost and isolation forest algorithms are used in the system's development on the SDN controller to monitor all traffic flows. The system uses a dynamic trust score engine to determine the traffic, which helps the system evaluate the traffic's reliability and implements the policy decision modules to enforce the policies. The Developed framework can detect several types of threats, namely port scanning, lateral movement, DOS, DDOS, and brute force attacks.