CAREER: An Integrated Scheduling Framework for Multicore Based Real-Time Embedded Systems

这个职业生涯项目的目标是设计和开发智能车辆中基于网络的多核控制系统的集成调度框架。现代汽车的先进功能(如稳定性控制和防撞)对其电子控制系统提出了很高的计算要求，该系统可以由多核处理器提供动力，以减轻其日益增长的复杂性。然而，现有的调度理论和技术已经不能支持这种基于网络的多核控制系统，特别是考虑到人的因素和智能车辆的动态环境，本项目对基于网络的多核智能车辆控制系统中的资源管理技术和调度算法进行了全面的研究，以有效地调度各种实时应用，有效地利用计算能力。首先，正在研究一种带有简化的高级主控制器的分级控制体系结构，以实现准确的态势感知并确保快速响应。其次，针对动态环境中的不确定性，开发了分布式、多核感知的弹性实时调度算法，该算法能够自适应地调整各种控制任务的调用间隔，以保证可调度性和稳定性。此外，正在开发关键程度感知调度算法，该算法考虑控制任务在不同情况下的不同重要性。最后，对混合动力/电动智能汽车的节能多核调度算法和电池管理方案进行了研究，提出的调度框架的设计和开发具有直接的经济和社会影响，可以减少车祸，从而挽救生命。此外，随着网络物理系统新课程的开发，该项目还为代表性不足的学生提供了丰富的主题和学习机会。

The goal of this CAREER project is to design and develop an integrated scheduling framework for networked multicore-based control systems in smart vehicles. The advanced functionalities (such as stability control and collision avoidance) in modern vehicles impose high computational demand on their electronic control systems, which can be powered by multicore processors to mitigate their increasing complexity. However, the existing scheduling theory and techniques have fallen short of supporting such networked multicore-based control systems, especially considering the human-related factors and dynamic environments of smart vehicles.This project undertakes a comprehensive study of resource management techniques and scheduling algorithms to efficiently schedule various real-time applications and effectively utilize the computation power in networked multicore-based smart vehicle control systems. First, a hierarchical control architecture with a simplified high-level master controller is being investigated to achieve accurate situational awareness and ensure prompt response. Second, to address the uncertainty in dynamic environment, distributed and multicore-aware elastic real-time scheduling algorithms are being developed that can adaptively adjust the invocation interval of various control tasks for schedulability and stability. Moreover, criticality-aware scheduling algorithms are being developed that consider the variable importance of control tasks under different situations. Finally, energy-efficient multicore scheduling algorithms and battery management schemes for hybrid/electric smart vehicles are being investigated.The design and development of the proposed scheduling framework has a direct economic and societal impact, which can reduce car accidents and thus save lives. Moreover, with the development of new curricula on cyber-physical systems, this project also provides abundant topics and learning opportunities for under-represented students.