【作者】 曾勇；

【导师】 马建峰；

【作者基本信息】 西安电子科技大学 ， 密码学， 2008， 博士

【摘要】 容忍入侵是整合容错与密码技术的一种安全防护方法,容忍入侵的目的不是设计出绝对安全可靠的系统,也不是限制或者阻止恶意攻击者对系统脆弱性的利用,而是设法保证:当系统某些部件已经受到破坏或者被控制时,系统仍然能够提供核心的或者基本的服务,并保护数据的机密性与完整性。近几年,容忍入侵以及相关领域方面的研究得到了广泛关注。然而,已有研究成果的应用要求相应的系统具有高计算能力和高存储能力,同时对系统冗余部件的数目与质量有较高要求,组件之间的通信也要求具有高数据率和低传输出错率,这使得容忍入侵的实现需要付出相对高昂的代价。另一方面,很多现实应用场景由于资源有限,从而不具备完全实现容忍入侵的条件。本文通过对现有研究成果的分析,指出了在资源受限场景下对基本服务的保护是容忍入侵方法研究的新方向,并且对这个方向下的一些关键技术进行了如下研究:1.讨论了资源受限场景下的可靠性区分容错路由机制。基于已有方案,对基于无状态机制的可靠性区分容错路由机制建立了等价数学优化模型,证明了该模型在多数应用场景下没有解析解,这表明现有方案的求解将带来不可忽略的计算开销,因而不适用于资源受限场景。鉴于此,提出了基于多路径概率负载均衡的改进方案,给出了启发式的轻量级求解算法,并用具体示例与仿真实验检验了算法的有效性;2.提出了基于三角不等式原理的限制性前向转发策略,该策略通过引入虚拟参考点,将数据的转发限制在特定区域从而减少了无关节点的能耗,通过与已有限制性数据转发机制的比较,表明了所提策略的优点;随后讨论了该区域的面积与可靠性的函数关系,得到了可靠性随着参考点与源节点距离的增长而平方级增长的结论,在此基础上给出了减少无关节点参与的可靠性区分容错路由机制,通过相关实验说明了算法的有效性;最后给出了适用于资源受限场景的整合加密与签名功能的容错数据传输方法;3.提出了网络入侵检测中的流量抽样理论分析模型。分析了大规模网络入侵检测中流量载荷的抽样分析技术,随后以微分方程为工具,对网络的流量变化进行了量化分析,结合基于流量分析的入侵检测技术,建立了网络异常检测的流量分析数学模型;进一步的,通过对某些类别的入侵行为的流量特性分析,结合提出的数学模型,给出了关于抽样检测间隔的上界,进而从理论上给出了对特定入侵行为来选择流量抽样间隔的方法;4.综合论述了仿生算法在密码学与信息安全中的一些有趣尝试性工作,分析了这些工作的优缺点,指出了仿生算法在密码学与信息安全中的四类研究方向,一、基于机器自动验证的安全协议设计方法;二、基于人工免疫的入侵检测系统设计;三、基于人工智能难题的低熵口令保护与密钥协商方法;四、演化算法的单向性分析以及基于此单向性的伪随机数生成。随后重点关注了基于演化计算的伪随机数生成方法,以及如何采用困难人工智能难题减轻客户端设备的计算与通信开销,从而为资源受限的客户端提供安全的服务。更多还原

【Abstract】 Intrusion-tolerance is the technique of using fault-tolerance to achieve security properties. Instead of focusing on intrusion prevention or eliminating all the system vulnerabilities, it should provide an ultimate defense of the system, the aim of which is to fulfill the primary missions in the presence of intrusion or partial compromising. The tolerance paradigm in security has deserved much attention recently.However, the available intrusion tolerant architectures have the following strict requirements: 1) the system should have high ability of computing and storage; 2) the redundant components of the system should satisfy some tight requirements about quantity and quality; 3) the communications among components require reliable and high speed transmission. As a result it takes comparative cost to implement general intrusion tolerance. On the other hand, many application scenarios have tight resource restriction. This dissertation presents that how to provide prime services in resource constraint scenarios is a new progress of intrusion tolerance. Then some key problems in this direction are studied:1. The fault-tolerant routing in resource constraint scenarios is discussed. Firstly, a mathematical optimal model for available fault-tolerant stateless routings is given. Then the model is proved no analytical solutions in many scenarios, which shows that available solutions have non-ignorable computing cost, as a result not suitable for resource constraint scenarios. This thesis presents an improved fault-tolerant routing with probability load balance among multiple paths, and a lightweight algorithm for the model. In the end, the efficiency of the model and the algorithm is shown by an illustration and simulation.2. Triangular inequality based forwarding strategy is proposed. Firstly, the data forwarding are restricted in a specified area by introducing a virtual reference point. Then the relationship between reliability and this area is given. A detailed analysis shows that the reliability increases quadratically with the distance between the source node and the reference node, based on which the reliability differentiated fault-tolerant routing algorithm is given. In the end, the advantages of proposed algorithm over the related work are shown by analysis and comparisons.3. A network traffic sampling model for the network intrusion detection system (NIDS) is given. Based on differential equation theory, a quantitative analysis of the effect of NIDS on the network traffic is given firstly. Secondly, a minimum delay time of NIDS needed to detect some kinds of intrusions is analyzed. In the end, an upper bound of the sampling distance is discussed. Proofs are given to show the efficiency of our approach.4. Recently some nature-inspired algorithms are used in cryptography and information security. This thesis shows the advantages and disadvantages of those interesting work. And four feasible directions are summarized: 1). automatically designing provably secure protocol; 2) designing artificial immunity and intrusion detection system; 3) password based authentication and key establishment using hard artificial problem; 4) one-way property of evolutionary algorithm. Then the thesis focuses on the pseudorandom number generation using one-way property of evolutionary algorithm. In the end, the thesis discusses how to reduce the resource cost in the client equipments when providing secure service.更多还原