软件定义网络的主动攻击与防御机制研究

软件定义网络（SDN）是一种新型计算机网络体系结构，具有中心式的网络控制、高效的数据转发和灵活的编程接口等特性，可极大的简化日趋复杂的网络管理任务。然而，正是由于数据/控制平面分离以中心控制的特点，SDN很容易受到主动攻击的威胁。考虑到SDN在国际学术和产业界的迅猛发展，以及在国民经济的广阔应用前景，对SDN的安全隐患特别是主动攻击进行系统的研究，并提出合理有效的防御机制显得十分必要和迫切。本项目从控制平面、控制信道、数据平面三个维度来分析SDN面临的主动攻击威胁，通过综合利用分布式哈希表、多队列调度、神经网络、同态消息验证码等技术手段，分别提出基于控制器策略完整性保障、异常OpenFlow消息检测和流量控制、轻量级的规则执行验证等研究方案，旨在对SDN中的策略篡改攻击、控制信道拒绝服务攻击、流表篡改攻击等主动攻击进行检测和防御，提高SDN网络的安全性、可信性和可靠性。

Software Defined Network (SDN) is a new network architecture, featured in centralized network control, highly-efficient data forwarding, and flexible programing interface. It is shown that SDN can greatly ease the increasing complicated task of network management. However, due to its decoupled control/data plane, and the centralized control mode, SDN is very vulnerable to many active attacks. Considering the worldwide attention from both the academia and industry, and the promising future in national economy, it is very urgent to systematically study the security vulnerabilities of SDN (especially active attacks faced by SDN), and come up with proper and effective counter-measures. This project approaches this problem from three aspects: control plane, control channel, and data plane. We propose a research plan including policy integrity guarantee for controllers, abnormal OpenFlow message detection and traffic control, and lightweight rule enforcement verification, by leveraging distributed hash table, multi-queue scheduling, neural network, and homomorphic MAC, etc. These security mechanisms can detect and prevent malicious modification of SDN policies, denial of service of control channel, and malicious modification of flow table, in order to enhance the security, trust, and reliability of SDN.