面向车载CAN网络的多维非功能属性集成优化算法  

Integrated Multi-dimensional Non-functional Properties Optimization Algorithm for Vehicle Controller Area Network

作　　者：谢勇[1] 陈旭辉[1] 吴武飞[2] 张佳[3] 

机构地区：[1]厦门理工学院计算机与信息工程学院,福建厦门361024 [2]湖南大学信息科学与工程学院,长沙410082 [3]集美大学计算机工程学院,福建厦门361021

出　　处：《计算机工程》2016年第3期29-33,52,共6页Computer Engineering

基　　金：国家自然科学基金资助项目(61502405;61300039;61173076);福建省教育厅中青年教师教育科研基金资助项目(JA15368);厦门市科技计划基金资助项目(3502Z20131158;3502Z20133033;3502Z20143031);厦门理工学院科技基金资助项目(YKJ13024R;XYK201437;YKJ15019R)

摘　　要：为解决车载CAN网络设计时没有同时考虑实时性、安全性和可靠性等多维非功能属性的问题,提出一种集成优化模型,将可扩展性和带宽利用率作为模型的优化目标。在满足实时性、可靠性和安全性定量要求的前提下,通过为可扩展性和带宽利用率增加权重的方式实现两者的统一优化,进而对不同目标进行归一化,使得在同一个值域内实现权衡优化。在此基础上,从信号打包问题入手,给出基于模拟退火的CAN系统多维非功能属性集成优化算法。在真实信号集和仿真信号集上的实验结果表明,与不考虑多维非功能属性的集成优化情况相比,该算法可获得较高的目标优化率。In order to solve the problem that multi-dimensional non-functional properties are not taken into consideration during the design of Controller Area Network（ CAN）,such as real-time,security and reliability,this paper proposes an integration optimization model. The scalability and bandwidth utilization ratio are used as the optimization objectives of the model. Under the premise of meeting the requirement of real-time,reliability and safety,by means of adding weights for scalability and bandwidth utilization,the unified optimization of both is realized. Then different objectives are normalized to get the balanced optimization within the range. On this basis,starting from the signal packing problem,it presents the multi-dimensional non-functional properties integration optimization algorithm in CAN system based on Simulated Annealing（ SA）. Experimental results on real signal set and simulated signal set showthat compared with the situation of integration optimization without considering multi-dimensional non-functional attributes,this algorithm can obtain higher objective optimization rate.

关 键 词：汽车电子系统 车载网络 多维非功能属性 信号打包问题 模拟退火 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]