众核片上混合关键系统安全高效资源管理研究

Mixed-Criticality Systems (MCS) on many-core platforms are an emerging area that is attracting more and more attention in the field of real-time systems. MCS on many-core platforms typically operate in hybrid, dynamic, and complex environments. These characteristics give rise to a large number of significant challenges facing safe and efficient resource access, and numerous problems call for urgent solutions.. Among many others, differentiated mixed-criticality service provision and limited-resource efficiency improvement are two important problems that remain to be addressed. With focus on these issues, this project will deal with safe and efficient resource management schemes that can be used in MCS on many-core platforms. Specifically, for safety-critical applications, their dynamic criticality requirement and workload will be modeled with multi-mode theory, and multi-mode mixed-criticality task mapping methods based on heuristic techniques will be proposed, which will be able to ensure mixed-criticality requirements and improve system performance and resource efficiency. For non-safety-critical applications, their dynamic utility will be modeled based on their significance and urgency, and auction-based admission control strategies will be devised, with capability of system utility optimization. Taking automobile electronic systems as application background, experimental platforms for resource management schemes of MCS on many-core platforms will also be developed and used for performance evaluation of the proposed solutions.. Expected outcomes will provide a scientific and theoretical basis for safe and efficient resource management of MCS on many-core platforms, and enable applications of MCS on many-core platforms with a complete and effective support.

众核片上混合关键系统（MCS）是实时系统领域新兴的研究热点。众核片上MCS具有高混杂、高动态和高复杂的特征，对安全高效的资源访问提出了许多新的挑战，很多问题亟待解决。. 本项目研究适用于众核片上MCS应用的安全高效资源管理关键技术，解决众核片上MCS中混合关键区分服务支持和有限资源高效利用的关键科学问题，实现安全高效的资源访问。针对安全关键应用，利用多模式理论对其动态关键性需求和工作负载进行建模，研究基于启发式的多模式混合关键任务映射方法，保证混合关键性需求并优化系统性能和资源效率；面向非安全关键应用，基于重要性和紧急性对其动态效用进行建模，建立基于拍卖的任务准入控制机制，优化系统效用；以汽车电子系统为应用背景，搭建众核片上MCS资源管理实验平台，进行实证研究。. 预期成果将为众核片上MCS的安全高效资源管理提供科学基础和理论依据，并为众核片上MCS的实际应用提供理论和技术支撑。

混合关键系统已经成为实时系统发展的必然趋势，其高混杂、高动态和高复杂的特征，对安全高效的资源管理提出了巨大挑战。针对这一问题和现状，本课题在混合关键系统资源管理及其应用方面开展了大量研究。项目执行期间，课题组共发表CCF-A类论文4篇，CCF-B类论文10篇，CCF-C类论文2篇，其他论文3篇，申请发明专利2项。课题组成员赴美国、法国和德国等参加国际学术会议7人次。项目主要研究成果包括：（1）在安全关键系统时间行为分析方面，提出了基于问题约减的任务链最坏延迟分析方法、基于概率性最坏情况执行时间整形的实时调度方法；（2）在概率性实时计算资源和通信资源管理方面，提出了面向限制截止期周期性实时任务的和谐划分调度方法、多处理器平台下固定优先级概率性实时任务的负载感知和谐任务映射方法、基于动态算法的通信资源管理方法；（3）在面向特定场景的实时调度应用研究方面，提出了面向无线网络化控制系统的概率性能耗感知实时调度方法、面向无线充电传感网的安全可靠实时调度方法、基于WiFi的高可靠低延时无设备身份验证方法、面向移动边缘计算的效用感知实时任务转移方法、面向物联联网的自适应实时数据采集方法、面向大数据的自动实时分类方法。以上研究工作的开展，突破了混合关键系统资源管理的一些基本理论问题，并且为其实际应用提供创新技术，能够推动汽车电子、航空航天、智能医疗等相关应用领域的发展。总之，课题组在项目执行期间按照既定年度计划和人员安排进行课题研究，并按时顺利完成了预期研究目标。