【作者】 张霞；

【导师】 詹瑾瑜；

【作者基本信息】 电子科技大学 ， 软件工程（专业学位）， 2014， 硕士

【摘要】 随着计算机信息网络的不断发展,基于网络的嵌入式实时系统也更多的应用于人类的生活生产之中。然而,嵌入式实时系统的网络化在带来极大便利的同时,随之而来的便是极大的挑战。嵌入式实时系统与互联网紧密相连,必然导致系统遭受来自网络的恶意攻击的可能性变大,尤其是近几年嵌入式系统与物联网的联系更加紧密,这意味着系统的失效可能导致严重的人生财产损失。这就要求系统引入安全保护服务以抵御这些安全威胁,然而安全保护的引入必然导致更多的系统负载,危及系统的实时性,从而加剧系统的失效风险。与此同时,嵌入式系统由于其应用环境的特殊性,其能耗供给往往非常有限,这就要求嵌入式系统在保证安全的同时严格遵守系统的能量约束。因此如何权衡系统的安全性,实时性及能耗有效性,为系统提供最大限度的安全保护,成为一大挑战。本文主要针对分布式的应用,从安全,能耗和实时三个方面,在任务调度层面对嵌入式实时系统进行设计优化。本文着眼静态任务调度,在系统设计阶段通过统筹的考虑系统中所存在的所有任务及所有可用的资源,探索安全性能加强,能量使用有效的嵌入式实时任务的调度策略。首先,对常用对称加密算法RC5及RC6算法的安全性能进行建模,通过分析已发布的测试数据及对算法本身的结构,对安全性能及相应的时间能耗开销进行量化。在此基础上,针对分布式的数据采集的应用,分析应用可能面临的安全问题及相应的安全需求,面向安全关键的独立周期任务,建立安全感知的静态任务调度模型,以安全性能加强和实时性确保为调度目标,寻求安全性能最优的任务调度算法,并就其有效性和局限性进行探讨。另外,针对分布式的安全及能耗关键嵌入式实时系统及分布式并行任务,分析其安全能耗需求,提出了安全加强,能量有效及实时性保证的静态调度模型。我们主要通过总线数据加密和总线通信量减少两种途径来提升系统安全性能,在此基础上寻求安全及能耗感知的静态任务调度方法。对安全敏感的混合关键实时系统进行分析和研究,抽象出基于系统行为的混合安全关键的系统模型,提出安全敏感的混合关键实时系统能量优化的问题,并在此基础上给出相应的调度方法。更多还原

【Abstract】 With the rapid development of computer network, more and more networked real-time embedded systems are introduced to human’s everyday life. There is no doubt that these systems make our life more convenient. However, close connection to the Internet results in higher possibility of being attacked by hackers. Especially as real-time embedded systems are increasingly applied in the Internet of Things in recent years, failures of the systems are more likely to cause casualties and property loss. So introduction of security service into these systems is necessary. But more security protection means more system load, which makes violation of deadline more possible and thus aggravates risk of system failure. At the same time, energy supply of embedded systems is usually limited. Thus, when strengthening security protection of embedded system, energy constraints must be taken into consideration. To sum up, researchers must tradeoff among system security, real-time requirement and energy efficiency to provide the systems with better protection. And this has become a big challenge.This thesis mainly concentrates in task-scheduling-level optimization of real-time embedded systems while security, energy and real-time of the system are all considered. Static task scheduling is the main topic. In the thesis, the major job is to seek efficient scheduling strategy for real-time tasks, which reinforce systems’ security protection and increase energy efficiency as well, by taking all tasks and available system resource into account.Firstly, the security of two widely used symmetric cryptographies RC5 and RC6 is modeled, and their security performance, corresponding time and energy consumption are quantified by analyzing publicized test data and structure of the cryptographies. Then the security requirements of distributed data collection systems are specified, and the static security-aware scheduling model for security-critical independent periodic tasks is established. After that a security-aware task scheduling algorithm is proposed to achieve the objection of increasing security protection within the constraint of real-time requirement. And finally its effectiveness and limits should be discussed. The thesis also addresses scheduling problems of parallel tasks in distributed security- and energy-critical real-time embedded systems. Firstly the security and energy requirements are analyzed and formulated, based on which the security optimization problem is proposed. And then a Genetic Algorithm based mechanism is derived to handle this problem. This mechanism increases the security protection through two ways, encryption of messages and communication reduction. In the last part of this thesis, the research focuses on security-sensitive mixed-critical real-time system. Firstly the behaviors of mixed security-critical system are formulated. Then the energy optimization problem of security-sensitive mixed-critical systems is proposed and a scheduling mechanism is derived to handle this problem.更多还原