随着现代电子信息技术的飞速发展和汽车制造业的强劲增长，现代汽车技术与电子技术、信息技术的融合使得汽车不再是简单的交通工具，而成为现代科技的载体和结晶。汽车电子的发展使得车内各种电子控制单元ECU（Electronic Control Unit）日益增多且日趋复杂，单纯靠几个功能模块完成各自的任务已经远远不能达到系统高性能、实时、安全、可靠的要求，这样就需要有相应的操作系统来控制和协调各个功能模块，从而保证了汽车能够安全可靠、高性能的运作。由于汽车对安全性、实时性等方面的要求比较高，同时针对汽车内的操作系统还有相应的标准要求，所以一些通用的操作系统并不适合汽车领域。因此，需要针对汽车领域不同的功能特点，研究特定应用的汽车专用操作系统vASOS（Vehicular Application Specific Operating Systems）。 上世纪90年代，王飞跃教授提出了特定应用操作系统ASOS（Application Specific Operating Systems）的概念，而汽车专用操作系统vASOS正是特定应用操作系统ASOS在汽车领域的延伸。本文在特定应用操作系统ASOS研究的基础上，利用实时系统相关的理论方法，从体系结构和调度算法两个方面深入研究了特定应用的汽车专用操作系统vASOS。本文的主要工作和贡献如下： 1、根据汽车内总线传输速度以及ECU单元处理对象的不同，将汽车内部系统分为三大类，高端ECU单元、低端ECU单元以及车载信息系统。从不同系统面向处理对象、实时性要求以及不同功能特点出发，根据ASOS应用特制微内核SCM（Specific Customized Micro-kernel）的思想建立了面向高端ECU、低端ECU以及车载信息系统的vASOS体系结构模型，指出了三大部分的相同点和不同点，同时利用统一建模语言UML建立了三部分的vASOS体系结构模型并对其进行了分析。 2、根据高端ECU下任务类型及调度的要求，提出了面向高端ECUvASOS系统的最优调度指标，并在对相关实时调度算法（空闲挪用算法和带宽保留算法）分析的基础上，提出了高端ECU vASOS的优化调度算法，即在满足所有周期性任务时间限的前提下，调度非周期性任务使任务在满足时间限的情况下，响应时间越小越好。 3、分析了低端ECU vASOS的任务类型及调度要求，以不精确计算的调度思想建立了低端ECU的任务模型，提出了面向低端ECU vASOS系统的最优调度算法，即：如果一个调度算法能够在满足1）周期性任务的强制部分是可调度的；2）服务于非周期任务的服务器的加入，不会影响系统中周期性任务的强制部分可调度性的前提下，使得周期性任务的可选部分的平均调度误差最小，那么该调度算法就是最优的。 4、服务于车载平台的车身操作系统vASOS不是一个单约束类型的实时系统，而是一个存在着多级实时约束和非实时应用并存的混合实时系统。本文在对相关调度算法（预留资源算法和基于服务质量QoS的调度算法）分析的基础上，建立了基于服务质量QoS的车载信息系统vASOS混合调度算法模型。在该模型中，硬实时任务使用经典的EDF算法，软实时任务采用了预留资源算法，同时针对预留资源算法的不足，提出了基于PID反馈的预留资源调度算法。在系统资源分配方面，提出了基于服务质量QoS的优化分配算法，并进行了实验仿真。
英文摘要
With rapid developments of advanced electronics technologies and vehicles manufacture industries, automobiles are employing a lot of modern technologies to improve their performance and reliability. However, developments above largely increased complexity of Electronic Control Units (ECU) in automotive applications. Therefore, it is difficulty to meet requirements of good performance, real-time and safety of vehicles any more by using several simple functional modules as before. Thus certain operating systems would be a better choice. But because of some special requirements in vehicle domain, some common operating systems are not suitable for vehicles. Therefore, aiming at different functional characteristics of vehicles, it is necessary to study Vehicular Application Specific Operating Systems (vASOS). Application Specific Operating Systems (ASOS) was originally proposed by Professor Fei-Yue Wang in early 1990s. vASOS is just the extension of ASOS in vehicle domain. Based on current researches of ASOS, this dissertation makes deep researches on vASOS from architecture to scheduling algorithms with theories of real-time systems. The main contributions of this dissertation include following issues: (1) According to differences of transmission speed of buses and manipulated objects of ECUs, we divide vehicular systems into three types: High-end ECU, Low-end ECU and In-vehicle information systems. Based on analysis of differences on manipulation objects and functional specific requirements, three different vASOS frameworks are designed with the idea of ASOS Specific Customized Micro-Kernel by using UML. (2) According to task models and scheduling requirements in High-end ECU, an optimal scheduling algorithm for High-end ECU vASOS is proposed, which not only guarantees all period tasks finished before their deadlines, but also minimizes response times of sporadic tasks. (3) Through analyzing Low-end vASOS task models and scheduling requirements, imprecise computation idea is used to design task models and an optimal scheduling algorithm is addressed to 1) guarantee scheduleability of mandatory parts of period tasks, 2) guarantee aperiodic tasks involved not influence the scheduleability of mandatory parts of period tasks, and 3) minimize the average scheduling error of optional components of period tasks. (4) In-vehicle information systems vASOS is a multi-constraints real-time and non-real-time application coexisted real-time system. To meet the system scheduling requirements of In-vehicle information systems vASOS, a QoS-based real-time scheduling framework is presented in this dissertation. In this mechanism, hard real-time tasks use EDF scheduling algorithm, while soft real-time tasks use reservation-based feedback scheduling algorithm. Meanwhile, an optimal QoS-based resource allocation algorithm is provided and the effectiveness of the method is verified by simulation.
修改评论