汽车工程 ›› 2023, Vol. 45 ›› Issue (9): 1563-1572.doi: 10.19562/j.chinasae.qcgc.2023.09.006

所属专题: 智能网联汽车技术专题-规划&决策2023年

• • 上一篇    下一篇

面向智能汽车的SOA架构及服务调度机制研究

郝建平,苏炎召(),钟志华,黄晋()   

  1. 清华大学车辆与运载学院,北京 100086
  • 收稿日期:2023-07-20 修回日期:2023-08-30 出版日期:2023-09-25 发布日期:2023-09-23
  • 通讯作者: 苏炎召,黄晋 E-mail:alexsu0916@tsinghua.edu.cn;huangjin@tsinghua.edu.cn
  • 基金资助:
    国家自然科学基金委联合基金重点项目(U20A20285);优秀青年基金项目(52122217)

Service-Oriented Architecture and Service Scheduling Mechanism for Intelligent Vehicles

Jianping Hao,Yanzhao Su(),Zhihua Zhong,Jin Huang()   

  1. School of Vehicle and Mobility,Tsinghua University,Beijing 100086
  • Received:2023-07-20 Revised:2023-08-30 Online:2023-09-25 Published:2023-09-23
  • Contact: Yanzhao Su,Jin Huang E-mail:alexsu0916@tsinghua.edu.cn;huangjin@tsinghua.edu.cn

摘要:

面向服务的架构(service-oriented architecture, SOA)是软件定义智能汽车的核心设计理念。本文对智能汽车SOA软硬件层级进行了分析和归纳,并对影响SOA调度的关键环境因素进行了分类。基于此,本文提出了可拓展功能域的SOA架构,建立了SOA服务层级和服务响应参数模型,设计了SOA服务调度机制。该SOA服务调度机制以服务置信度为核心,综合考虑SOA子服务的服务能力、预计运行时间、负载能耗和运行稳定性等影响因素,实现了当前环境系统约束下的子服务最优选择。在高速路、拥堵道路、低光照隧道场景下,本文采用简化的双层SOA模型进行了模拟仿真验证,结果表明:与不进行服务调度方法相比,本文所提出的SOA架构及调度方法能够在保持相近的服务能力下, 降低计算资源及能耗负载约36%,并降低约30%的软件预计服务时间。

关键词: 面向服务的架构(SOA), 智能汽车, 服务调度, 服务置信度, 服务质量

Abstract:

Service-Oriented Architecture (SOA) stands as the core design concept for software-defined intelligent vehicles. This paper presents an analysis and synthesis of the software and hardware layers of SOA in intelligent vehicles, along with a categorization of key environmental factors influencing SOA scheduling. Building upon this analysis, this paper proposes an extended-domain SOA architecture tailored for vehicles, establishes a model for SOA service layers and service response parameters, and devise a service scheduling mechanism for this architecture. The service scheduling mechanism centers around the concept of service confidence, and takes into account multiple factors such as service capability, estimated runtime, resource consumption, and performance stability of sub-services, to make optimal sub-service selections under the current environmental constraints. Simulation tests are performed using a simplified double-layer SOA model in scenarios including highways, congested roadways, and low-light tunnels. Results indicate that compared to scenarios without service scheduling, the proposed SOA architecture with the scheduling mechanism significantly reduces computing load and energy consumption by approximately 36% and decreases the predicted service time by around 30%, all while maintaining comparable service capacity.

Key words: SOA, intelligent vehicles, service scheduling, service confidence, quality of service