在車用隨意網路中以交易代幣為基礎之叢集化方法

Abstract

由於車用隨意網路的發展，駕駛人已經可以透過車上裝置與其他車輛或是路側裝置分享路況資訊，作為行車時選擇道路的參考依據。在市區或是重要幹道中，由於車輛密度較高，若是單純使用廣播方式傳遞路況資訊，容易造成廣播風暴問題，而建立叢集(Cluster)成為了一種常見的解決方式。 一般的叢集化方法都希望最小化管理叢集的負擔，但此類方法在叢集化時只參考車輛ID或連接數，在VANET (Vehicular Ad-Hoc Network)中仍有改進空間。本研究以最少叢集改變(Least Cluster Change, LCC)演算法為基礎，提出了交易代幣式叢集化(Token-Based Clustering, TBC)演算法，將車輛地理位置與資訊更新時間數值化成代幣(Token)，並讓車輛藉由代幣交易機制購買路況資訊，讓擁有最多代幣的車輛將成為叢集管理者(Cluster Head)。車輛付出代幣向臨近的車輛購買路況資訊，並依買方擁有的代幣數量與賣方資訊的更新時間計算交易價格。由於處於重要地理位置或已擁有較新路況資訊的車輛容易累積較多代幣，所以適合成為叢集管理者。 本研究以細胞自動機(Cellular Automata)建立交通環境模型，並比較在車輛密度與比例不同時，使用LCC演算法與TBC演算法的差異。在實驗中，發現TBC演算法在車輛密度與比例較高時，較能降低環境中的叢集數量，讓叢集中的成員數提高，並減少訊息傳遞的長度。

With the development of vehicular ad-hoc network (VANET), drivers can share traffic information through inter-vehicle communication and take traffic information into their driving consideration. Because of the higher vehicle density in urban area, broadcasting traffic information to other vehicles will cause the broadcast storm problem. For the problem, clustering is a common solution. Reducing the overhead in cluster management is important for general clustering methods. Base on least cluster change (LCC) algorithm, the study propose a token-based clustering (TBC) algorithm. Being different from existing clustering algorithm, in our model vehicles pay tokens to buy traffic information and the price of the information is calculated based on update time and vehicle location. The vehicles at important location or with latest traffic information are easier to accumulate tokens, so the vehicles are more suitable to be cluster heads than the other vehicles. The study use cellular automata (CA) to construct a simulation traffic environment and analyzes different cases with varied traffic densities and proportion of vehicles. Experimental results show that TBC algorithm can reduce the number of cluster and decrease the number of packet hops.