連通性觸發之即時交通資訊車際網路路由

Abstract

車際網路是近年來新興的研究議題。利用車際網路上的車間通訊開發的各種應用，可以幫助人們生活更加便利且安全。這些應用要得以實現，都需要依靠路由機制讓資料可以在兩車輛間順利傳送。在路由機制的設計上，先前已經有很多研究出現，各種選擇最佳路徑時的考慮因素，如道路上的車子數量，道路長度等都已被提出。本文提出一個利用即時交通資訊的路由協定，藉由路段上位於路口的車輛發動探測封包，經由路段上的車輛攜帶和傳送到達同一路段上的另一路口，由此測量得到各條路段上最新的封包傳送延遲時間，並且考慮車輛在道路路口轉彎時會減速，車輛數量較多的特性，在不加入額外路邊單元的情況下，設計機制將收集到的延遲時間資訊暫時儲存在路口範圍內，同時嘗試將此資訊擴散到鄰近路口並暫存。當封包經由車輛轉傳到路口，需要對下一個即將轉傳的路段作優先權計算時，便可得到此資訊，在沒有得到即時延遲資訊的路段，因為車輛上有預先紀錄的歷史交通資料(車輛數，道路速限)，車輛可以依據先前研究中的算式估算各路段的傳輸延遲時間，混合有即時延遲時間資訊路段的和經由歷史資料計算出延遲的路段，得到最短延遲時間的路徑。進一步考慮更新即時交通資訊所帶來的大量網路負擔，我們研究各個路段的連通性，先判斷路段是否為接近連通，也就是使用無線傳輸可以快速傳送資料經過此路段，中途不會因傳輸中斷而需要由車輛攜帶，這些路段對傳送延遲時間的影響較大，針對這些路段進行即時資訊更新的動作，藉以減少網路負擔量同時增進傳輸延遲時間。連通性的考慮方法為利用車輛在行駛的特性，當車輛自由行駛時，道路上車輛的分佈呈現指數分佈。道路上每隔一段長度內有車輛的機率可以用道路上的車密度得到。於是只要在傳輸範圍內存在有一台以上車輛出現的機率極高時，便代表可以往前傳送。連續數個傳輸範圍內都有車輛可以幫忙轉傳，傳輸的總長度大於道路長度時，道路接近於完全連通。但考慮車輛在道路上的分佈並非均勻且實際每次往前轉傳時轉傳距離不等於傳輸範圍，必須針對轉傳時的轉傳距離做計算，最後以連續數個期望轉傳距離內皆有車輛的機率當作連通機率，可求出車輛密度和連通機率的關係。 透過模擬比較和其他路由協定的傳輸延遲時間，傳輸成功率和協定本身對網路造成的負擔。在正常交通情況下結果略好，但當交通情況出現和統計資料有重大差異時，本協定可以增進封包傳輸的延遲時間。

VANET is a popular research topic in recent years. Along with inter-vehicle communication in VANET, a number of applications have been developed to help people’s lives. To make these applications come true, we need a routing protocol to deliver packets between vehicles. In this paper, we propose a routing protocol using real-time traffic information. Vehicles on intersections are triggered to send connectivity packets to vehicles on adjacent intersections conditionally. These packets traverse road segments in two ways, forwarding and carrying, and allow us to measure the latest road delivery delay. We consider the feature of vehicles slowing down and gathering on intersections, so the number of cars is more there. We try to design a mechanism to store delay information on intersection area without Road Side Unit (RSU). Meanwhile, we will disseminate it to cars on neighbor intersections and store on them in the same way. When packets are forwarded to vehicles on one intersection, the routing protocol will calculate road priority and assign packets to vehicles on the road with higher priority as possible. In this step, if one vehicle can get real-time delay information, we can use it. Otherwise we use statistical density data to estimate road delay for roads without real-time information as in previous works. It mixes these two kinds of road delay sources to decide the next forwarding road. Furthermore, we take periodical update overhead into account. We discuss road connectivity probability and find out the relation between this probability and the number of cars on the roads. When there’s a higher connectivity probability on a road, it means packets can be forwarded through road segment by wireless transmission without carrying, thus the delay is shorter. During the traversal process of connectivity packets, we judge whether a road segment is connected. If a road is connected, then we update road delay of the road. These roads have significant impact on end-to-end delay, so we focus and do update action on them. Then we can have the benefit of reducing overhead and improving delay at the same time. To research connectivity, we assume the distribution of inter-distance of cars is exponential when drivers go on road freely, referencing traffic research. For one car, if there is another one within its wireless transmission range, it can forward packets to that car and packets can move ahead to destination intersection for the expected forwarding distance. We use average vehicle spacing to find the expected forwarding distance. From the start of one intersection, we calculate the probability that a vehicle exists within the transmission range of one car. When this situation happens many times, the total forwarding distance exceeds the road length. We say the road is connected and use this process to find out the connectivity probability. When we sense that the connectivity probability is high enough, we send the connectivity packets. Through simulations, we find out that our method works and has better performance in some cases. In traffic congestion cases, simulation result shows that our method can improve delay and reduce overhead compared with other routing protocols.