Software-Defined Firewall: Enabling Malware Traffic Detection and Programmable Security Control.

Shang Gao,Zecheng Li,Yuan Yao,Bin Xiao,Songtao Guo,Yuanyuan Yang

DOI: https://doi.org/10.1145/3196494.3196519

2018-01-01

Abstract:Network-based malware has posed serious threats to the security of host machines. When malware adopts a private TCP/IP stack for communications, personal and network firewalls may fail to identify the malicious traffic. Current firewall policies do not have a convenient update mechanism, which makes the malicious traffic detection difficult. In this paper, we propose Software-Defined Firewall (SDF), a new security design to protect host machines and enable programmable security policy control by abstracting the firewall architecture into control and data planes. The control plane strengthens the easy security control policy update, as in the SDN (Software-Defined Networking) architecture. The difference is that it further collects host information to provide application-level traffic control and improve the malicious traffic detection accuracy. The data plane accommodates all incoming/outgoing network traffic in a network hardware to avoid malware bypassing it. The design of SDF is easy to be implemented and deployed in today's network. We implement a prototype of SDF and evaluate its performance in real-world experiments. Experimental results show that SDF can successfully monitor all network traffic (i.e., no traffic bypassing) and improves the accuracy of malicious traffic identification. Two examples of use cases indicate that SDF provides easier and more flexible solutions to today's host security problems than current firewalls.