验证信息全携式域间路由安全机制研究

In existing digital signature based schemes for securing interdomain routing, there are still some pending problems on security. Its design is restricted by the pressure of bandwidth of update messages and overhead of on-line computations. For improving the safety, three security problems are managed to be solved, which are transient faults causing long-term failures for ROA (Route Origin Attestation) access path, attacks related to repositories, and customer RPKI CA prefix hijacking. For solving the first security problem, ROAs are put in update messages so as to eliminate relying circularities. For solving the second security problem, information related to authenticate correctness of public keys is also put in update messages, so as to eliminate repositories which are attacking points. In these two solving methods, all necessary authenticating information is carried in update messages, which will cause increase of pressure of bandwidth of update messages and overhead of on-line computations. Therefore, in order to lower pressure of bandwidth of update messages while overhead of on-line computations are considered, new methods are presented for address attestations and route attestations creating and verifying. For solving the third security problem, distincts between RPKI CA prefix hijacking and MOAS (Multiple Origin Autonomous Systems) are made explict by the randomness of choosing of transient private keys. Formal derivation and verification, which are to verify safety of proposed scheme, are managed to be presented based on safety criterion and the assumption of abilities of attackers.

现有基于数字签名的域间路由安全机制在安全性上还有一系列问题有待解决，其设计受到路由通告带宽压力和在线计算量的制约。为提高安全性，拟解决其中的三个安全问题：路由起源声明(ROA)发布点访问的临时出错可能导致前缀地址块长期不可达,针对存储池发布点的攻击,下层认证中心(RPKI CA)前缀劫持。对于安全问题一，将ROA置于路由通告内部发送，以消除循环依赖。对于安全问题二，将公钥正确性证明信息置于路由通告内部发送，以消除存储池发布点，使攻击者没有攻击点。这两个解决方法要求验证信息全携带于路由通告中，会导致路由通告带宽压力和在线计算量的增加。为此，构造新的地址证明和路径证明创建和验证方法，以降低路由通告带宽压力，同时兼顾在线计算量。对于安全问题三，利用临时私钥选择的随机性，使下层CA前缀劫持与多源自治系统(MOAS)间的区别显性化。拟基于安全准则和攻击者能力假设，对机制安全性进行形式化推导证明。

现有基于数字签名的域间路由安全机制在安全性上还有一系列问题有待解决，其设计受到路由通告带宽压力和在线计算量的制约。为提高安全性，解决其中的三个安全问题：路由起源声明(ROA)发布点访问的临时出错可能导致前缀地址块长期不可达,针对存储池发布点的攻击,下层认证中心(RPKI CA)前缀劫持。对于安全问题一，将ROA置于路由通告内部发送，以消除循环依赖。对于安全问题二，将公钥正确性证明信息置于路由通告内部发送，以消除存储池发布点，使攻击者没有攻击点。这两个解决方法要求验证信息全携带于路由通告中，会导致路由通告带宽压力和在线计算量的增加。为此，构造了新的地址证明和路径证明创建和验证方法，以降低路由通告带宽压力，同时兼顾在线计算量。对于安全问题三，利用临时私钥选择的随机性，使下层CA前缀劫持与多源自治系统(MOAS)间的区别显性化。. 在提高安全性的同时，为降低路由通告带宽压力，将传统基于组织域名的方式转换成基于IP前缀号和自治系统号的方式，并采用Nyberg-Rueppel聚合签名方案，整体进行带内验证时路由通告带宽增量为2000 Bytes,远低于路由通告4096 Bytes的容量上限。使路由通告验证所需数据可在路由通告内部传输(带内)。为降低在线计算量，采用Rabin签名结合Nyberg-Rueppel聚合签名方案，使域间路径验证过程中的在线计算量大幅度降低。可望提高域间路由安全机制的安全性、降低在线计算量和路由通告带宽压力，促进部署实施。