- About this Journal
- Abstracting and Indexing
- Aims and Scope
- Annual Issues
- Article Processing Charges
- Articles in Press
- Author Guidelines
- Bibliographic Information
- Citations to this Journal
- Contact Information
- Editorial Board
- Editorial Workflow
- Free eTOC Alerts
- Publication Ethics
- Reviewers Acknowledgment
- Submit a Manuscript
- Subscription Information
- Table of Contents
International Journal of Distributed Sensor Networks
Volume 2013 (2013), Article ID 837381, 10 pages
Driving Path Predication Based Routing Protocol in Vehicular Ad hoc Networks
1Yunnan Key Laboratory of Computer Technology Applications, Kunming University of Science and Technology, Kunming, Yunan 650500, China
2Faculty of Civil Engineering and Architecture, Kunming University of Science and Technology, Kunming, Yunan 650500, China
Received 9 November 2012; Accepted 16 January 2013
Academic Editor: Chao Song
Copyright © 2013 Yong Feng et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The vehicular mobility is reflection and extension of the human social activity. Since human trajectories show a high degree of temporal and spatial regularity, thus vehicular driving paths are predictable to a large extent. In this paper, we firstly analyze the predictabilities of different types of vehicles and then propose a new driving path predication based routing protocol (DPPR). With hello messages to broadcast vehicles’ driving path predication information to neighbor vehicles, DPPR can observably increase the successful ratio to find the proper next hop vehicles that move toward the optimal expected road in intersection areas. In roads with sparse vehicle density, DPPR utilizes vehicles to carry messages to roads with high vehicle density while the messages’ forward paths partially coincide with the vehicles’ driving paths. Moreover, as to messages that can tolerate long delay, they can be carried to destinations by vehicles whose driving paths will pass the messages’ destination in order to optimize bandwidth utilization. Simulation results demonstrate the effectiveness of the proposed DPPR protocol.
In recent years, vehicular networks have gradually become an important research field in wireless communication and computer network and received broad attention from both industry and academy [1, 2]. As vehicle nodes’ high speed mobility and uneven distribution, vehicular networks have dynamic and changing network topology, which makes it difficult to maintain persistent connection among vehicle nodes. To improve data delivery performance, vehicular networks widely adopt ad hoc networking and delay tolerant networking technology. Thus vehicular ad hoc networks (VANETs), which evolved from mobile ad hoc networks (MANETs) and delay tolerant networks (DTNs) and are formed by cars and any supporting fixed nodes, emerges and are regarded as an important component of intelligent transportation systems (ITS) .
VANETs promise a wide range of valuable applications including real-time traffic estimation for trip planning, mobile access to Internet, and in-time dissemination of emergency information such as accidents and pavement collapses. To realize the applications above, one of the key research topics is to design effective and efficient data delivery schemes. Therefore, many schemes have been presented to solve the problem in recent years. Among the existing schemes, some works mainly take advantage of geographic position information, such as GPSR  and CAR . The performances of these protocols mainly depend on the network connectivity, and they are sensitive to the vehicle node density. Many works are based on the traffic statistics and network layout, such as SADV  and VADD . A few protocols such as TBD , TSF , and STDFS  are designed to utilize available vehicle trajectories to improve the data delivery performance. To disseminate vehicle trajectory information, these protocols assume that numerous wireless access points (APs) need to be deployed along roads. That will undoubtedly request a large amount of investment. Some researchers have made explorations on the prediction of vehicle driving paths and take use of the character of anticipating vehicle routes to develop data delivery schemes. In the literature [11–14], several prediction models and schemes are proposed. However, these works are based on either historic driving records or local and current vehicle running status, and thus there are prediction accurate problems.
Many research works [15–17] show that human trajectories show a high degree of temporal and spatial regularity: each individual can be characterized by a time-independent characteristic length scale and a significant probability to return to a few highly frequented locations. Sincerely the vehicular mobility is reflection and extension of the human social activity; thus vehicular driving paths are predictable to a large extent. In general, vehicles can be categorized to three types: (i) bus, tramway, and light rail, which have stable trajectories and schedules; (ii) private car, which has remarkable regular trajectories and obviously temporal and spatial regularity; for example, a private car generally travels among limited several places such as home, workplace, supermarket, and park; (iii) taxi, which has flexible and variable running paths. In this paper, we analyze the driving path predictability for different sorts of vehicles and discuss the corresponding prediction methods. Doing this makes ordinary vehicles foresee their driving paths. Based on the character, we propose a new driving path prediction routing protocol (DPPR), which can increase data delivery ratio through (i) carrying messages by vehicles whose driving paths will pass through the messages’ destination; (ii) increasing chance to find the vehicles that move toward the optimal expected road in intersection areas. The major contributions of this work may be listed as follows. (a)We analyze the driving path predictability of various types of vehicles with a microscopic perspective and divide vehicles into three categories in which the prediction method of each category vehicle is discussed. Moreover, we also discuss the feasibility and practicability of utilizing the vehicle driving path prediction method to improve the routing performance in VANETs.(b)We propose a new routing protocol called DPPR to improve the data forwarding performance of VANETs, which effectively takes advantage of the characteristic of foregone driving path for average vehicles. Through extensive simulations, the effectiveness of our proposed DPPR scheme is evaluated.
The rest of the paper is organized as follows: Section 2 summarizes the related works of routing protocols in vehicular ad hoc networks. In Section 3, we analyze the driving path predictability of various types of vehicles and discuss the feasibility and practicability of DPPR scheme. In Section 4, we describe the design of our proposed DPPR routing protocol in detail. Section 5 shows the effectiveness of DPPR via simulation experiments. Section 6 concludes the paper.
2. Related Works
In recent years, data delivery and forwarding issues about vehicle-to-vehicle and vehicle-to-infrastructure in VANET have gained lots of attention [1–6]. For the frequent network partition and merging due to the high mobility of vehicles, the physically constrained nodal mobility resulted from the fixed roadways, and for the constrained nodal moving speed limited by the roadway conditions, the data forwarding in VANET is different from that in the traditional mobile ad hoc networks (MANETs). These unique characteristics of the road networks make the MANET routing protocols ineffective in the VANET settings . Thus, many routing protocols based on carry-and-forward thinking have been proposed in order to reach efficient and effective data forward performance in VANETs.
Among these works, Epidemic  is an early approach to deal with the data forward issue in frequent network partition and merging settings. It allows the random pair wise exchange of data packets among mobile nodes in order to maximize the possibility that data packets can be delivered to their destination node. Thus, a great number of copies of data packets are generated during the delivery, which weakens its performance to a great extent, especially when the resources of bandwidth and buffer are limited.
Some works mainly take advantage of geographic position information, such as GPSR , CAR , MMR , and VVR . Similar to GPSR, both MMR and VVR use greedy forwarding strategy to find the next packet carrier based on the geographical proximity toward the packet destination. Through using the approach of “guard node,” CAR forwards data packets through the connected path from the packet source to the packet destination. The performances of these protocols mainly depend on the network connectivity, and they are sensitive to the vehicle node density. Thus the geographic position based routing schemes cannot work well when the vehicular traffic is sparse and of noneuniform-distribution.
Some works are based on the traffic statistics and network layout, such as SADV , VADD , and DBR . SADV routing leverages on the stationary nodes to improve the network connectivity and the data forward performance. Through using a stochastic model based on vehicular traffic statistics, VADD tries to achieve as high delivery successful ratio as possible with low delivery delay from mobile vehicles to stationary packet destinations. DBR scheme focuses on satisfying the user-defined delay bound rather than the lowest delivery delay, so that it can economize resource such as channel utilization and buffer space.
A few protocols such as TBD , TSF , and STDFS  are designed to utilize available vehicle trajectories to improve the data delivery performance. TBD utilizes the vehicle trajectory information along with vehicular traffic statistics in order to compute the accurate expected delivery delay for better forwarding decision making. However, to disseminate vehicle trajectory information, these protocols assume that numerous wireless access points (APs) need to be deployed along roads. That will undoubtedly request a large amount of investment.
Some researchers have made explorations on the prediction of vehicle driving paths and take use of the character of anticipating vehicle routes to develop data delivery schemes. Several prediction models and schemes are proposed such as PBR , MOPR , and PLR . PBR exploits the location and velocity information of vehicles to predict route lifetime and takes preemptive action to minimize route failure. MOPR improves the routing process by selecting the most stable route in terms of lifetime. PLR uses a location predictor to solve the problem of location inaccuracy and vehicle mobility. In , Jeung et al. propose a network mobility model to predict the driving paths of vehicles. However, these works are based on either historic driving records or local and current vehicle running status, and thus there are prediction accurate problems. Moreover, the approaches mainly focus on short-term and short-distance prediction and do not take into account how long-distance driving path prediction information is used to improve the data forwarding in VANETs.
3. Vehicle Driving Path Prediction
In this section, we firstly review the mobility trajectory features of various types of vehicles and analyze the predictability of their driving paths with a microscopic perspective. According to the differences of predictability, vehicles are classified as three categories in which the prediction method of each category vehicle is discussed. Moreover, we also discuss the feasibility and practicability of utilizing the vehicle driving path prediction method to improve the routing performance in VANETs.
3.1. Driving Trajectory Predictability for Vehicles
In the real world, there exist various kinds of vehicles. As to the stability of driving route, the vehicles can be divided into three categories: (i) the first sort of vehicles that have certain driving trajectories, such as bus, tramway, and light rail, which have stable trajectories and schedules; (ii) the second sort of vehicles that have remarkable regular trajectories, such as private car whose trajectory has obviously temporal and spatial regularity; for example, a private car generally travels among limited several places such as home, workplace, supermarket, and park; (iii) the third sort of vehicles that have changing trajectories, such as taxi, which has flexible and variable running paths. Based on the above classification, we will respectively discuss the driving path predictability for each sort of vehicles as follows.(a)It is clear that the driving paths of the first sort of vehicles are fully predictable; as a result of this feature they have stable trajectories and schedules.(b)The driving paths of the second sort are predictable to a large degree because the vehicles have relative regular trajectories and their driving destination places are generally limited and relatively fixed. Some works have focused on the research topic of vehicle driving path prediction, and a few prediction models and methods have been proposed in recent years. For example, in , Liu et al. develop a heuristic and context-dependent induction method based on decision trees, to predict vehicle moving trajectories. PLR  uses a location predictor to solve the problem of location inaccuracy and vehicle mobility. In , Jeung et al. propose a network mobility model to predict driving paths of vehicles.(c)As to the third sort of vehicles, it seems difficult to predict their driving paths because the vehicles have flexible and variable running paths. Taking a deeper look at the trajectory feature of the third sort of vehicles, however, we find that the prediction difficulty results from the fact that the vehicles’ destination locations are uncertain, that is, their destinations may be any place. But if their destination information can be collected in the initial stage of every trip, then the sort vehicles’ trajectories will be predictable in a large part. In our previous work , we discuss the issue of destination information gathering and propose a driving path prediction method for taxis based on destination information gathering. For the third sort of vehicles, they will always choose the paths with the shortest driving time to reach their destinations as soon as possible. Based on the gathered destination information, GPS device, electronic map, and traffic statistic information at different times, a practicable prediction method is to utilize the Dijkstra algorithm to look for the lowest cost path from the current position to the destination, where the cost means the average travel time for each road segment. To further improve prediction accuracy, we will take advantage of the feature that drivers are relatively fixed and familiar with the road and traffic conditions. Through recording the drives’ historical route and personal preferences at different times on the on-board unit of each vehicle itself, the prediction path can be adjusted and more consistent with the real situation. In general, it is reasonable of presume that the driving paths third sort of the vehicles’ are predictable while corresponding prediction methods are assisted by destination information.
From resource requirement aspect, necessary hardware equipments and software systems may include on-board unit, GPS device, electronic map, traffic statistic, input device such as phonetic recognition device, handwriting board, keyboard, and touch panel. Because the above equipments and systems are already popular and becoming common in vehicles, there is hardly any hardware, software, and cost problem to predict vehicles’ driving paths.
From prediction scheme aspect, for the first sort of vehicles that have stable trajectories and schedules, undoubtedly, driving path prediction is fully feasible. As to the second sort, as they have relative regular trajectories and stable destinations, a few meritorious prediction methods and models have already been proposed and discussed, and thus it is feasible to predict their driving trajectories with a high reliability. To predict the trajectories of the third vehicles, the first step is to collect vehicle destination information. As previously discussed, there is no technology and cost problem because whether phonetic recognition or handwriting board, keyboard, and touch panel they are all very mature and reliable technology. The key problem lies to create an incentive measure that can prompt drivers to carry out the information gathering activities. A proper solution can be that senders or receivers should pay a fee to drivers for messages’ success delivery.
From the analyses above, we can get such a useful feature from most vehicles, that is, each vehicle can know the driving path beforehand or in the beginning of its current trip. Through broadcasting driving path information to neighbor vehicles in hello messages, each vehicle can know the trajectories of the vehicles that it will meet on road, which will bring such benefits for improving routing performance in VANETs as follows.(a)To increase data delivery success ratio through carrying messages to their destinations by vehicles whose driving paths will pass the messages’ destination. The carrying message method is very effective in regions where vehicle density is sparse and especially useful for applications that can tolerate long delivery delay due to high delivery successful ratio and negligible low transmission cost.(b)To improve multiple-hop forwarding success ratio. For time-sensitive applications, multiple-hop forwarding is essential, but the difficulty lies in how to find vehicles that move toward the optimal expected road in intersection areas. Fortunately, driving path information beforehand will effectively solve the problem and increase the chance to find proper next hop.(c)To reduce wireless channel resources occupancy. For it is envisioned that a wide variety of applications can be running on VANETs in the near future, ranging from road safety, cooperative driver, to entertainment and Internet access, the coexistence of a large spectrum of vehicular applications means that they will inevitably compete with each other for the use of finite wireless network resources . That will easily lead to severe congestion at these areas with high vehicle density and likely result in the appearance of “hotspot” in the rush hour. Therefore, the carrying method can lessen wireless collision possibility and improve the quality of channel, which in turn can improve the delivery success rate.(d)Effective vehicle trajectory forecast can provide new opportunity to forward data from static locations to mobile vehicles, even between two or more mobile vehicles. In VANETs, existing routing protocols usually only consider how mobile vehicles send data message to location-fixed destinations, but seldom involve how to send messages to specific mobile vehicles, much less two-way data transmission between two moving vehicles. Obtaining driving path information, however, will make it possible or easier to realize message forwarding from fixed positions to mobile vehicles and two-way data transmission between moving vehicles.
4. Driving Path Predication Based Routing Framework
Through utilizing the proposed driving path prediction method in Section 3, each vehicle can know its driving path in advance. Therefore, a new valuable character is introduced into VANETs, that is, the driving trajectories of vehicle nodes are of foreknowledge. Based on this characteristic, we propose a driving path predication based routing protocol called DPPR to improve the data delivery performance in VANETs. In the rest of this section, our proposed DPPR will be described in detail. In Figure 1, the sketch map of DPPR thinking is given out.
Every vehicle can obtain its current location through GPS device and can be equipped with a preloaded street-level digital map, which not only describes road topology and traffic light period but also provides traffic statistics such as traffic density and average vehicle speed on roads at different times of the day. Such kind of digital map has already been commercialized , and more detailed traffic statistics will be integrated into digital map in the near future. Vehicles communicate with each other through short range wireless channel and can find their neighbors through beacon messages. Each beacon provides vehicle’s information such as its unique ID, location, velocity, and direction. What more, and each vehicle will carry out the prediction method proposed in Section 3 and announce its driving paths in beacon message.
More formally, the street-level digital map is abstracted as a directed graph . For any two intersections and , if and only if there is a road segment connecting and , and vehicles can travel from towards on that segment. And the notation used in the paper is listed in Table 1.
4.2. Path Selection Algorithm
In VANETs, a vehicle needs to send messages in three cases as follows: (a) the vehicle itself produces messages; (b) it forwards the message which it received from other vehicles; (c) it periodically takes from its routing buffer to send. When vehicle has a message to send, it needs to leverage the knowledge of global traffic statistics such as average vehicle density and speed on the road segments of the directed graph . Based on the information, DPPR works out the path with optimized performance for . As shown in Figure 2, vehicle near intersection wants to send a message to the park near intersection for reserving a parking space. The first important issue is to select an optimal forwarding path to deliver the message among the three paths: , , and .
For each edge in , we will evaluate it with transmission delay metric, that is, the time cost used to transmit a message through the edge. To estimate the transmission delay of a road segment (e.g., ), we utilize the delay model proposed in , which is shown in (1) as follows:
Since such information as road segment distance, vehicle density, and average velocity is available, we can get the transmission delay of each road segment through (1), so that each edge has a weight, that is, its transmission delay, in the directed graph . The proposed DPPR protocol assumes that the best path to the destination is the shortest delay one, that is, the path that minimizes the sum of transmission delay of the edges on the directed graph that abstracts the street map. By running a single invocation of Dijkstra on , the shortest delay path can be worked out.
4.2.1. Straightway Mode
As previously mentioned, each vehicle maintains a neighbor list by periodically broadcasting beacons. According to the content of the beacon messages, each vehicle can get the driving path information of its neighbors. In DPPR, as to a vehicle, its driving path is defined as the intersection sequence along which it will arrive at the destination of its current trip. According to changing geographical position, DPPR protocol will switch between Straightway and Intersection modes. In this section, we present the protocol used in the Straightway mode.
Since the traffic is at most bidirectional, data forwarding in the Straightway mode is simple. As vehicle has a message to send, it firstly operates the path selection algorithm to determine the nearest target intersection then chooses the optimal next hop vehicle from its neighbor list, and the chosen one is indicated as Nexthop. In the Straightway mode, we can simply apply the geographically greedy forwarding to choose Nexthop. At the same time, the vehicle also looks for the vehicles whose driving paths will pass the destination of the message. If there is more than a matching vehicle, then the one with the fastest average running speed is selected out, and its vehicle identification is denoted as Dest_accord. As shown in Algorithm 1, after completing the above query steps, vehicle can make a forwarding decision as follows.
Case 1. Driving path of itself will pass M’s destination: if Dest_accord is found and has faster average running speed than , then is sent to Dest_accord. Otherwise, checks whether there exists Nexthop; if so, it generates a copy of M, denoted as , increments flag_copy of , and sends it to Nexthop.
Case 2. Driving path of itself will not pass M’s destination: if Dest_accord is found and the distance between and message M’s destination is within the range of two intersections, then is sent to Dest_accord. Otherwise, checks whether there exists Nexthop; if so, then is sent to Nexthop.
When no proper Nexthop or Dest_accord is found, vehicle only puts into its routing queue. While receiving a message from another vehicle, the current firstly checks whether it has to keep a copy of the message in its routing queue. If so, it directly drops the message, else or puts the message into its queue.
4.2.2. Intersection Mode
When a vehicle enters an intersection, it switches to Intersection mode. For each packet in its buffers, the vehicle checks whether there is a contact available to forward the packet to the next road segment along the selected optimal path. If so, the vehicle transmits the packet to the neighboring vehicle, otherwise it continues to carry it. For example, when vehicle has a message to send, it firstly operates the path selection algorithm to sort all the outgoing directions. Secondly, looks for the neighbor vehicles whose running directions are in accordance with the optimal direction based on their driving path information. If there is more than a matching vehicle, then the one which is geographically closest towards the optimal direction is selected out, and its vehicle identification is denoted as Direction_accord. But if such Direction_accord cannot be found, will apply the geographically greedy forwarding to choose Nexthop, that is, the vehicle which is the nearest to the optimal direction among all its neighbors. As shown in Algorithm 2, after completing the above query steps, vehicle can make a forwarding decision as follows.
Case 1. Running direction of itself is the same as the optimal: if Direction_accord is found and is geographically closer towards the optimal direction than , then is sent to Direction_accord. Otherwise, puts into its routing queue and goes on carrying it.
Case 2. Running direction of itself is not the same as the optimal: if Direction_accord is found, then is sent to Direction_accord. Otherwise, checks whether there exists Nexthop; if so then is sent to Nexthop. When neither Direction_accord or Nexthop is found, this means that has no proper forwarding contact at present; it only puts into its routing queue.
4.3. Queue Management Algorithm
For each mobile vehicle node in VANETs, the size of its routing buffer queue is limited. Therefore, the queue management algorithm would greatly influence the data delivery performance. In the proposed DPPR, the flag_copy of a message indicates how many copies have been propagated, and the survival time shows how long a certain message has existed in the network. Therefore the flag_copy and survival time together denote the importance of a message, and the queue management is just based on the two parameters.
Messages are sorted in the routing queue based on an increasing order of their flag_copy. For those messages with the same flag_copy value, they are further sorted according to an increasing order of survival time. Thus messages with smaller tickets and shorter survival time are closer to the top of the queue, and can be transmitted with higher priorities. Moreover, messages will be dropped in the following two occasions: (a) when a message arrives and the queue is full, it is compared with that message at the end of the queue and the one with bigger flag_copy is dropped among them. If the flag_copy values of the two messages are equal, then the one with a longer survival time is dropped; (b) whenever a message’s survival time is longer than the delay tolerance of the network, it is dropped to avoid unnecessary resource occupation.
5. Performance Evaluations
In this section, we will evaluate the impact of the proposed DPPR scheme on the data transmission performance in VANETs. We choose the classic wireless ad hoc network routing algorithm GPSR  and Epidemic as the referential. Since pure GPSR has no carry-and-forward ability, which will result in poor performance in intermittently connected VANETs, we extend it by adding buffers and thus make it have basic carry-and-forward ability. In the following simulation experiments, the data delivery performances of GPSR only with simple carry-and-forward ability, denoted as GPSR (with buffer), Epidemic, and the proposed DPPR will be analyzed and evaluated.
5.1. Simulation Setting
The experiment is based on a 4000 m × 3000 m approximate rectangle street area, which is derived and normalized from a real street map of Beijing city in China. The rectangle contains 25 intersections and 40 bidirectional roads. We utilize NS-2.34 as the simulation tools. Since modeling of complex vehicle movement is important for accurately evaluating protocols, the open source software VanetMobiSim-1.1  is used to generate the movement of vehicles. Detailed simulation parameters are shown in Table 2.
To evaluate the performance of the protocols in different traffic density environments, we deploy two kinds of different number of vehicles, that is, 100 and 300, respectively, into the network to imitate different vehicle node density, which means different network connectivity. Vehicles are with the average speed ranges from 40 to 80 kilometers per hour. Among all vehicles, 20 of them are randomly picked out to send CBR data packet to fixed spots. To evaluate the performance on different load status, we change the CBR rate from 0.1 to 1 packet per second.
5.2. The Packet Delivery Ratio
In this section, we compare the performance of DPPR protocols with GPSR (with buffer) and Epidemic routing in the aspect of packet delivery ratio. Here, the packet delivery ratio is defined as the fraction of packets that have reached their destination without exceeding the delay tolerant threshold of certain application. For multiple-copy routing algorithm such as Epidemic and DPPR, a message is regarded as one successfully delivered only if any one copy of all its copies is received by the destination. We vary the data generation rate, that is, from 0.01 to 0.1 message/s, in order to evaluate the data delivery performance of the three protocols under different transmission loads.
Figures 3 and 4 show the data delivery ratio as a function of the CBR sending rate under two different vehicle densities; that is, the total number of vehicles is 100 and 300, respectively. From the two figures, we can find that the data delivery ratios of all three protocols go down with the increasing data sending rate, but the decreasing tendencies are much different. Epidemic routing has best performance and almost reaches the upper bound of the data delivery ratio when the data sending rate is very low; however its performance quickly deteriorates with the increase of sending rate. This is due to MAC layer collision and rapid exhaustion of the limited resources (i.e., network bandwidth, routing buffer, etc.) resulted from forwarding a tremendous amount of copies in epidemic routing. Though GPSR is extended and thus has simple carry-and-forward ability, its performance is still the worst in the case of sparsely connected networks, as shown in Figure 3. However, its delivery performance is very steady and almost insusceptible to the change of data sending rate. From the figure, we also can find that our proposed DPPR algorithm outperforms the others as the data sending rate increases, especially when the data sending rate is normal or high.
Figure 4 shows the data delivery ratio of the three protocols under high vehicle density environment that is, when the total vehicle number is 300. We can find that the change tendencies of data delivery ratio among these protocols remain unchanged as the data sending rate increases. As the increment of vehicle density improves the network connectivity, the performance of DPPR and GPSR routing is obviously enhanced. particularly for GPSR, it achieves relatively good delivery ratio when vehicle number is 300 because the network connectivity is improved with the increasing vehicle density. On the other hand, the high vehicle density is not good news for Epidemic protocol. The reason is that the number of message copies increases dramatically in epidemic routing as the node density increases, which results in an increasing number of collisions and dropped data packets.
5.3. The Packet Delivery Delay
In this section, we compare the packet delivery delay among DPPR, Epidemic routing, and GPSR protocols when transmitting data from moving vehicles to fixed destination spots. Figures 5 and 6 show the packet delivery delay as a function of the CBR sending rate under two vehicle densities; that is, the total number of vehicles is 100 and 300, respectively. From Figure 5, we can see Epidemic routing has the least delivery delay when the data sending is very low, but its performance drastically deteriorates; that is, its delivery delay rapidly goes up, as the data sending rate increases. GPSR has relatively low data delivery delay at low node density, but it is not meaningful simply because of its low delivery ratio. We also see the delivery delay of DPPR algorithm is generally lower than that of GPSR and Epidemic, and it has much lower delay than Epidemic especially while data sending ratio is high.
Figure 6 shows data delivery delay of the three protocols when the total vehicle number is 300. We can find that the delay of epidemic routing dramatically increases with the increasing data sending rate because it generates a great number of redundant packets. As the traffic load increases, many packets may be dropped, which is similar to the case in Figure 5. The delivery delay of DPPR and GPSR shows slowly decreasing tendency. As the increment of vehicle density improves the network connectivity, we also see that the delay of DPPR and GPSR is clearly lower than that in Figure 5, and the proposed DPPR routing shows optimal performance in general.
In near future, traffic safety and many other commercial applications will be running on VANETs. For realizing these promising applications, it is an important research topic to develop effective routing protocols that can reach high data delivery performance. In this paper, we analyze the driving path predictability for different sorts of vehicles and discuss the corresponding prediction methods. Doing this makes ordinary vehicles foresee their driving paths. Through utilizing the features, we propose a new routing protocol called DPPR to improve the performance of data delivery in VANETs. Through simulation experiments, we analyze and evaluate the performance of GPSR (with buffer), Epidemic, and DPPR protocol. The experiment result shows that our proposed DPPR protocol can reach high data delivery performance in VANETs.
This work is supported by the National Natural Science Foundation of China under Grant no. 61262081; the Yunnan Provincial Applied Fundamental Research Project under Grant no. KKSY201203027.
- J. Ott and D. Kutscher, “Drive-thru internet: IEEE 802.lib for “automobile” users,” in Proceedings of the IEEE INFOCOM Conference on Computer Communications, pp. 362–373, March 2004.
- D. Reichardt, M. Miglietta, L. Moretti, P. Morsink, and W. Schulz, “Cartalk 2000—safe and comfortable driving based upon inter-vehicle communication,” in Proceedings of the IEEE Intelligent Vehicles Symposium (IV '02), pp. 545–550, Jun 2002, http://www.cartalk2000.net.
- Research and I. T. A. (RITA), “IntelliDrive: Safer, Smarter and Greener,” http://connection.ebscohost.com/c/articles/52707131/intellidrive-safer-smarter-greener.
- B. Karp and H. T. Kung, “GPSR: greedy perimeter stateless routing for wireless networks,” in Proceedings of the 6th Annual International Conference on Mobile Computing and Networking (MobiCom '00), pp. 243–254, ACM, August 2000.
- V. Naumov and T. R. Gross, “Connectivity-aware routing (CAR) in vehicular ad hoc networks,” in Proceedings of the 26th IEEE International Conference on Computer Communications (INFOCOM '07), pp. 1919–1927, May 2007.
- Y. Ding, C. Wang, and L. Xiao, “A static-node assisted adaptive routing protocol in vehicular networks,” in Proceedings of the 4th ACM International Workshop on Vehicular Ad Hoc Networks (VANET '07), pp. 59–68, September 2007.
- J. Zhao and G. Cao, “VADD: vehicle-assisted data delivery in vehicular Ad hoc networks,” IEEE Transactions on Vehicular Technology, vol. 57, no. 3, pp. 1910–1922, 2008.
- J. Jeong, S. Guo, Y. Gu, T. He, and D. Du, “TBD: trajectory-based data forwarding for light-traffic vehicular networks,” in Proceedings of the 29th International Conference on Distributed Computing Systems (ICDCS '09), Montreal, Quebec, Canada, June 2009.
- J. Jeong, S. Guo, Y. Gu, T. He, and D. H. C. Du, “TSF: trajectory-based statistical forwarding for infrastructure-to-vehicle data delivery in vehicular networks,” in Proceedings of the 30th IEEE International Conference on Distributed Computing Systems (ICDCS '10), pp. 557–566, June 2010.
- F. Xu, S. Guo, J. Jeong et al., “Utilizing shared vehicle trajectories for data forwarding in vehicular networks,” in Proceedings of the 30th IEEE International Conference on Computer Communications (INFOCOM '11), pp. 441–445, April 2011.
- V. Namboodiri and L. Gaw, “Prediction based routing for Vehicular ad hoc networks,” in Proceedings of the Vehicular Ad Hoc Networks (VANET '04), 2004.
- H. Menouarand, M. Lenardi, and F. Filali, “Movement Prediction-based Routing (MOPR) concept for position-based routing in vehicular networks,” in Proceedings of the IEEE 66th Vehicular Technology Conference (VTC '07), pp. 2101–2105, October 2007.
- Z. Li, Y. Zhu, and M. Li, “Practical location-based routing in vehicular ad hoc networks,” in Proceedings of the IEEE 6th International Conference on Mobile Adhoc and Sensor Systems (MASS '09), October 2009.
- H. Jeung, M. L. Yiu, X. Zhou, and C. S. Jensen, “Path prediction and predictive range querying in road network databases,” VLDB Journal, vol. 19, no. 4, pp. 585–602, 2010.
- M. C. González, C. A. Hidalgo, and A.-L. Barabási, “Understanding individual human mobility patterns,” Nature, vol. 453, no. 7196, pp. 779–782, 2008.
- S. Schönfelder and U. Samaga, “Where do you want to go today?-More observations on daily mobility,” in Proceedings of the 3rd Swiss Transport Research Conference, Ascona, Switzerland, 2003.
- K. Kelley, “Casual carpooling-enhanced,” Journal of Public Transportation, vol. 10, no. 4, p. 119, 2007.
- Y. Lee, H. Lee, N. Choi, Y. Choi, and T. Kwon, “Macro-level and Micro-level Routing (MMR) for urban vehicular ad hoc networks,” in Proceedings of the 50th Annual IEEE Global Telecommunications Conference (GLOBECOM '07), pp. 715–719, November 2007.
- A. Vahdat and D. Becker, “Epidemic routing for partially-connected Ad Hoc networks,” Technical Report, 2000, http://issg.cs.duke.edu/epidemic/epidemic.pdf.
- H. Lee, Y. Lee, T. Kwon, and Y. Choi, “Virtual Vertex Routing (VVR) for course-based vehicular ad hoc networks,” in Proceedings of the IEEE Wireless Communications and Networking Conference (WCNC '07), pp. 4408–4413, March 2007.
- N. Liu, M. Liu, J. Cao, G. Chen, and W. Lou, “When transportation meets communication: V2P over VANETs,” in Proceedings of the IEEE 30th International Conference on Distributed Computing Systems (ICDCS '10), 2010.
- Y. Feng, Q. Qian, F. Wang, and X. Fu, “Public transportation assisted data delivery scheme in vehicular delay tolerant networks,” in Proceedings of the 8th Mobile Ad hoc and Sensor Networks, Chengdu, China, 2012.
- A. Skordylis and N. Trigoni, “Delay-bounded routing in vehicular ad-hoc networks,” in Proceedings of the 9th ACM International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc '08), pp. 341–350, May 2008.
- “Digital Map Data,” 2012, http://www.mapmechanics.com/digital-map-data/.
- J. Härri, F. Filali, C. Bonnet, and M. Fiore, “VanetMobiSim: generating realistic mobility patterns for VANETs,” in Proceedings of the 3rd ACM International Workshop on Vehicular Ad Hoc Networks (VANET '6), pp. 96–97, ACM, September 2006.