Balancing Bandwidth Utilization and Interrupts: Two Heuristic Algorithms for the Optimized Design of Automotive CPS
Balancing Bandwidth Utilization and Interrupts: Two Heuristic Algorithms for the Optimized Design of Automotive CPS
复制标题
平衡带宽利用率和中断：汽车 CPS 优化设计的两种启发式算法
DOI：
10.1109/tii.2019.2936240
复制
发表时间：
2020-04-01
影响因子：
12.3
通讯作者：
Takada, Hiroaki
中科院分区：
文献类型：
--
作者：
Xie, Yong;Zeng, Gang;Takada, Hiroaki
要实现汽车网络物理系统的优化设计，需要考虑通信和计算之间的相互作用。但是，现有关于具有灵活数据速率（FD）的控制器区域网络（CAN）的研究忽略了这一点，它只考虑了带宽利用率的最小化，并忽略了它会触发太多不必要的消息接收中断的事实（MRI）关于接收电子控制单元的消息。为了解决这个问题，本文提出了一个权衡问题，该问题可以平衡带宽利用率和CAN FD设计过程中不必要的MRIS数量。我们首先提出了一种算法，以分析包装邮件触发的不必要的MRIS数量，然后提出了两种启发式算法，即自上而下的方法和混合方法，以解决中型和大型信号集的权衡问题， 分别。实验结果表明，与最先进的算法相比，自上而下的方法将不必要的MRIS减少了10.48＆x0025; -99.89＆x0025;只有0.02＆x0025; -1.32＆x0025;带宽利用率开销，混合方法将不必要的MRIS减少23.15＆x0025; -99.63＆x0025;只有0.13＆x0025; -2.07＆x0025;带宽利用率开销。
To realize the optimized design of the automotive cyber-physical system, it is required to consider the interplay between the communication and computation. However, the existing research about the design of the controller area network (CAN) with flexible data rate (CAN FD) ignores this, it only considers the minimization of the bandwidth utilization and neglects the fact that it would trigger too many unnecessary message receiving interrupts (MRIs) on message receiving electronic control units. To address this problem, this article formulates a tradeoff problem that balances the bandwidth utilization and the number of unnecessary MRIs during the design of the CAN FD. We first propose an algorithm to analyze the number of unnecessary MRIs triggered by the packed messages, and then, two heuristic algorithms, namely, the Top–Down approach and the Hybrid approach, are proposed to resolve the tradeoff problem for midsized and large signal sets, respectively. The experiment results show that compared with the state-of-the-art algorithm, the Top–Down approach reduces the unnecessary MRIs by 10.48%–99.89% with only 0.02%–1.32% bandwidth utilization overhead, the Hybrid approach reduces the unnecessary MRIs by 23.15%–99.63% with only 0.13%–2.07% bandwidth utilization overhead.