適應式隨插即用車輛系統之軟體選擇

Software Selection for Adaptive Plug-and-Play Automotive System

傅家靖(Chia-Ching Fu)

指導教授 ： 林忠緯

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2020年)

https://doi.org/10.6342/NTU202002826

摘要

隨著智慧型汽車的蓬勃發展，構成車輛的車內軟體也隨之增加。但我們發現環境狀況會對智慧型汽車車內軟體的效能有所影響，例如影像辨識與攻擊偵測軟體會因為環境變化而帶來不同的效能。另外一個問題是，有些車內軟體由於功能相容性的因素，是不能或必須被同時指派的，我們稱此問題為軟體相容性限制。因為上述兩個問題，我們應用了服務品質的數值來確保每個在特定環境下之任務可被服務品質達到要求的軟體所解決。並且，為了解決軟體相容性限制的問題，我們建立了軟體限制表來確保每次的軟體選擇皆不違反軟體相容性限制。為了在系統軟體中選擇適當的軟體來服務每個任務，我們引用了加權貪婪集合覆蓋演算法來選取所需要之軟體並以速率單調排程指派優先權得到最初解。最後，我們運用降火演算法在最初解上得到能產生總等待時間更小的軟體選擇與優先權排序的更優解。

關鍵字

智慧型汽車 ； 服務品質 ； 加權貪婪集合覆蓋演算法 ； 降火演算法 ； 總回應時間

並列摘要

As the rapid growth in the intelligent vehicles field, software programs required to compose a vehicle increase. However, we notice that environmental conditions can affect performances of software programs such as object detection and intrusion detection. Moreover, some software programs cannot or should be scheduled together due to software program compatibility constraints caused by software program functionalities compatibility. Therefore, to solve the above mentioned problems, the Quality of Service (QoS) value is used to guarantee each runnable under certain environment condition is being served by software programs with the QoS value that is larger or equal to that defined on the runnable, and the software program constraint table is constructed to guarantee software program selections do not violate software program compatibility constraints. To obtain the software program selection among all software programs in the vehicular system, we apply the Weighted Greedy Set Cover algorithm to sift through software programs. Afterward, rate-monotonic scheduling is applied to assign priorities to all selected software programs, from which the initial solution is obtained. At last, we apply the Simulated Annealing algorithm on the initial solution for a better software program selection and priority assignment that lead to a lower total response time.

並列關鍵字

Intelligent Vehicle ； Quality of Service ； Weighted Greedy Set Cover Algorithm ； Simulated Annealing Algorithm ； Total Response Time

參考文獻

[1] R. I. Davis, A. Burns, R. J. Bril, and J. J. Lukkien, “Controller area net-work (can) schedulability analysis: Refuted, revisited and revised,”Real-TimeSystems, vol. 35, pp. 239–272, 2007.

[2] S. Durga, S. Mohan, J. Dinesh, and A. Aneena, “Cuckoo based resource al-location for mobile cloud environments,” inComputational Intelligence, CyberSecurity and Computational Models, pp. 543–550. Springer, 2015.

[3] X. Fu, B. Liang, Y. Huang, X. Ding, and J. Paisley, “Lightweight pyramidnetworks for image deraining,” inIEEE Transactions on Neural Networks andLearning Systems, pp. 1794–1807. IEEE, 2018.

[4] B. Kim and S.-K. Park, “Ecu software updating scenario using ota technol-ogy through mobile communication network,” in2018 IEEE 3rd InternationalConference on Communication and Information Systems (ICCIS), pp. 67–72.IEEE, 2018.

[5] C.-W. Lin, B. Kim, and S. Shiraishi, “Hardware virtualization and task allo-cation for plug-and-play automotive systems,”IEEE Design Test, pp. 1–1,2019.