異質車載雲端計算系統之資源分配問題

Abstract

近來隨著相關研究與技術的進展，已存在許多車載網路的改良配置方式，而如何善用車載網路中有限的資源以此提升系統的服務品質成為近年來一個重要的課題。車載雲端計算(Vehicular Cloud Computing; VCC)的概念隨著此課題的延伸而漸漸成形， 在VCC系統中，其加入邊緣計算的概念並適當地整合車載霧端(由車子的計算資源組合而成)和遠程雲端，以提供使用者即時的霧端和雲端服務，包含計算、通訊和儲存資源。雖然過去研究已建立了VCC系統的資源分配模型，且以半馬可夫決策過程來獲得此模型的最佳資源分配策略。然而，過去研究卻很少考慮到異質車、路邊基地台(Roadside Unit, RSU)。因此，本研究提出一個半馬可夫決策過程模型，以此解決考量了異質車和RSU的VCC系統的資源分配問題，並求解其最佳資源分配策略。其中，異質車即為不同製造商製造的不同種類的車子，其配備將會因該車款檔次而有所不同，高檔車一般而言會有較好的車上配備而擁有較多的計算資源；而RSU不再只是作為通訊和傳輸資料之用，亦能用於VCC中的計算，可大為改善過去VCC系統中資源分配易出現短缺或是明明有足夠多的資源卻礙於有限的資源而無法分配最大量資源的情形。此外，當RSU資源未被使用時，也能成為VCC系統的常駐備用資源，亦即本強化了VCC系統的計算容量，使得VCC系統內能更分配的計算資源不易枯竭。模擬結果顯示本研究所提半馬可夫決策過程模型可精確刻畫VCC系統之資源分配，且在不同參數設定下均可獲得最佳資源分配策略。

Recently, there are many improved configurations of vehicle network with the progress of related researches and technologies, and how to use the limited resources in the vehicle network to improve the service quality of the system has become an important issue. The concept of vehicular cloud computing (VCC) system has been built gradually. VCC system coordinates the vehicular fog (consisting of vehicles’ computing resources) and the remote cloud properly to provide in-time services to users. Although pervious works had established the models for resource allocation in the VCC system based on semi-Markov decision processes (SMDP), few of them discussed heterogeneity of vehicles and influences of roadside units (RSUs). Heterogeneous vehicles made by different manufacturers may be equipped with different amount of computing resources; and furthermore, RSU can enhance the computing capability of VCC. Therefore, this work proposes an SMDP model for VCC resource allocation that additionally considers heterogeneous vehicles and RSUs, and an approach for finding the optimal strategy of VCC resource allocation. Simulation shows that the resource allocation in the VCC system can be captured by the proposed model, which performs well in terms of long-term expected values (consisting of consumption costs of power and time), under various parameter settings.