The BGP-based inter-domain routing system plays an important role in the Internet. Not only the performance of data forwarding, but also the topology, robustness, and security of the Internet severely rely on the inter-domain routing system. However, the BGP has some design flaws, which result in many serious security problems for the inter-domain routing system. Compared to other traditional attacks, such as Prefix Hijack, the Large-scale LDoS (LLDoS) attack against inter-domain routing system has a strong covert, which is reflected in its attack traffic and reactions caused by the attack are legal. The concealment of such attacks makes the existing security technologies difficult to deal with. This project would like to enhance the robustness of inter-domain routing system against LLDoS attacks from the perspective of survivability theory. It focuses on some issues including LDoS attacks detection, local inter-domain routing recovery, survival situation awareness of global inter-domain routing system and the emergent recovery of global inter-domain routes. On the basis of using CRF theory to merge the traffic characteristics and the context information of LDoS attack, the LDoS attack classification model for inter-domain routing nodes will be constructed. Instead of pre-computing failover paths, it will carry out local inter-domain routing recovery by combining the method of backup AS advertisement and reactive re-computing feasible paths. The global survivability status of inter-domain routing system will be assessed by quantifying the deviation extent of Update packets’ optimal feature from its normal state. At last, based on the centralized network control theory and the scale-free characteristic of inter-domain routing system structure, the global-consistent structured backup subgraphs will be generated. The research of this project is expected to make some breakthroughs in the related aspects, and provide theoretical and technical support for building survivable inter-domain routing system.
针对域间路由系统的大规模LDoS攻击具有很强的隐蔽性,其攻击流量和攻击引起的反应都是合法的,使得现有域间路由系统安全技术难以有效应对。本课题拟从生存性理论的角度出发,探索应对大规模LDoS攻击的解决方法,重点研究域间路由节点LDoS攻击检测、域间本地路由恢复、域间路由系统全局生存态势感知和域间全局路由应急恢复等生存性增强关键问题,拟运用条件随机场融合利用LDoS攻击的流量特征和上下文信息,建立域间路由节点LDoS攻击分类模型;综合基于先验式预计算备份路径的路由恢复方法和基于反应式按需计算路径的路由恢复方法等,提高域间本地路由恢复的效能;通过量化更新报文最优特征偏离正常态的程度来评估域间路由系统的全局生存态势;基于中心化控制的思想并结合域间路由系统结构的无标度特性,构造具有全局一致性的结构化备份子图。通过这些研究,期望在相关问题上有所突破,为构建安全可靠的域间路由系统提供理论和技术上的支持。
作为互联网的关键基础设施,基于BGP的域间路由系统的安全性对全局网络的健康稳定运行具有重要影响。然而,近年来针对域间路由系统的网络攻击技术日益复杂,尤其是利用BGP协议自适应机制特性的大规模LDoS(简称BGP-LDoS)攻击。其具有很强的隐蔽性和破坏性,使得现有域间路由安全技术难以有效应对。本课题从生存性理论的角度出发,探索应对大规模BGP-LDoS攻击的解决方法,重点研究了大规模BGP-LDoS攻击扩散机理、大规模BGP-LDoS攻击预防、大规模BGP-LDoS攻击检测、域间路由系统生存态势感知和域间路由应急恢复等方面内容。经过三年努力,在BGP-LDoS攻击扩散机理方面,提出了基于生灭过程的域间路由系统级联失效模型,并在此基础上,提出了基于级联失效模型的域间路由节点重要度评估方法;在BGP-LDoS攻击预防方面,提出了基于动态变换的域间路由系统BGP-LDoS攻击预防方法;在大规模BGP-LDoS攻击检测方面,提出了面向自治域的基于流量分类模型的BGP-LDoS攻击检测方法、面向全局的基于突变平衡态理论的BGP-LDoS攻击检测方法;在域间路由系统生存态势感知方面,提出了基于加权相似度的域间路由系统安全状态评估方法和基于多特征自适应融合的数据平面威胁感知方法;在域间路由恢复方面,提出了基于结构特性和中心化控制的域间全局路由应急恢复方法。另外,在BGP-LDoS攻击遏制方面,提出了基于协同过滤的BGP-LDoS攻击遏制方法。部分相关结果已形成学术论文和专利,在“Computer Networks”、《中国科学:信息科学》和《软件学报》等国内外重要学术期刊和会议上发表论文13篇,其中SCI、EI检索论文10篇。培养毕业博士研究生1名,资助出站博士后1名、硕士研究生2名;资助申请专利6项;课题负责人郭毅获得2016年军队科技进步一等奖1项,于2017年11月在清华大学计算机科学与技术博士后流动站完成博士后研究工作。
{{i.achievement_title}}
数据更新时间:2023-05-31
基于分形L系统的水稻根系建模方法研究
基于 Kronecker 压缩感知的宽带 MIMO 雷达高分辨三维成像
拥堵路网交通流均衡分配模型
低轨卫星通信信道分配策略
宁南山区植被恢复模式对土壤主要酶活性、微生物多样性及土壤养分的影响
面向RED路由器队列的LDoS攻击检测和防御方法的研究
域间路由系统的级联失效攻击检测与安全趋势预测方法研究
关联故障下的互联网域间路由系统健壮性研究
新一代互联网域间路由系统协同机理研究