多维可重构嵌入式平台动态资源管理及存储安全研究

Since storage wall, power wall, algorithm wall and other issues have become the performance bottleneck of embedded system, the implementation of adaptive task partitioning and resource reconfiguration with variable granularity are considered as new ways for enhancing computing efficiency of embedded system. Meanwhile, the research on security of embedded system is lagging far behind that of computer system or Internet. This research plans to study the confidentiality and integrity of data storage on the basis of establishing embedded security computing environment. Employs multi-level nesting keys to avoid system cost and latency associated with counter overflow and re-encryption of data block in confidentiality protection. Adopts non-balanced verification tree and dynamic allocation of data block to shorten the average check path length of integrity verification and improve verification efficiency, providing security protection for data storage without affecting the performance of the system. Meanwhile, investigates abstract models for reconfigurable task resources in different dimensions and granularity, which completely and accurately reflect the real property task and system. Applies 3D dynamic building block strategy and multi-dimensional reconstruction function combined method to provide efficient scheduling and resource management for improving system performance. Finally, designs and develops prototype machine of the whole system aiming at establishing theoretical and practical foundations for security protection mechanism of embedded system and efficient task calculation.

存储墙、功耗墙、算法墙等阻碍着系统计算效能提升，实现不同层次上、粒度可变的动态自适应任务划分和资源重构计算是提升计算效能的新途径；同时，嵌入式系统的安全性研究远远落后于通用计算机系统和互联网的安全性研究。项目拟在搭建嵌入式安全计算环境基础上，进行存储机密性与完整性研究。采用多级秘钥嵌套方法，解决机密性保护中因计数器溢出、秘钥与数据块更新带来的系统开销和延时问题；采用非平衡验证树与节点数据块大小动态分配方法，缩短平均校验路径长度，提高验证效率，在尽量不影响系统性能的前提下为其提供信息存储安全防护。研究不同维度、可变粒度任务资源的抽象模型，全面准确的反映出任务和系统的真实属性；采用三维动态积木策略与多维重构函数相结合的方法，为系统提供高效的任务资源自适应调度管理策略，提升系统整体计算效能；设计开发系统原理样机，为嵌入式系统安全防护机制的设计实现及任务的高效能计算执行奠定理论和实践基础。

存储墙、功耗墙、算法墙等阻碍着系统计算效能提升，实现不同层次上、粒度可变的动态自适应任务划分和资源重构计算是提升计算效能的新途径。本项目在搭建嵌入式安全计算环境基础上，进行存储机密性与完整性研究。采用多级秘钥嵌套方法，解决了机密性保护中因计数器溢出、秘钥与数据块更新带来的系统开销和延时问题；采用非平衡验证树与节点数据块大小动态分配方法，缩短平均校验路径长度，提高验证效率，在尽量不影响系统性能的前提下为其提供信息存储安全防护。研究了不同维度、可变粒度任务资源的抽象模型，全面准确的反映出任务和系统的真实属性；采用三维动态积木策略与多维重构函数相结合的方法，为系统提供高效的任务资源自适应调度管理策略，提升系统整体计算效能；针对云平台中传统方法无法动态部署虚拟机的问题，提出了一种基于时间约束的双目标虚拟机放置算法；最后设计开发系统原理样机，为嵌入式系统安全防护机制的设计实现及任务的高效能计算执行奠定理论和实践基础。