You are on page 1of 14

www.ietdl.

org
Published in IET Communications Received on 8th March 2009 Revised on 20th July 2009 doi: 10.1049/iet-com.2009.0163

In Special Issue on Vehicular Ad Hoc and Sensor Networks

ISSN 1751-8628

Real-time vehicle route guidance using vehicle-to-vehicle communication


J.-W. Ding1 C.-F. Wang2 F.-H. Meng1 T.-Y. Wu3
Department of Information Management, National Kaohsiung University of Applied Sciences, Taiwan Department of Computer Science, National PingTung University of Education, Taiwan 3 Department of Electrical Engineering, Tamkang University, Taiwan E-mail: jwding@cc.kuas.edu.tw
2 1

Abstract: With advances in wireless communications and global position system-enabled devices (such as cellular phones, PDAs and car PCs), vehicle route guidance systems gradually become indispensable equipments for more and more automobile drivers because of its great convenience. Conventional route guidance systems are designed to direct a vehicle along the shortest path from the origin to the destination. However, usually, the shortest path does not result in the smallest travel time because of the dynamic trafc conditions on roadways. Therefore the state-of-the-art route guidance systems incorporate real-time trafc information to nd better paths. There are two types of approaches to collecting real-time trafc information: infrastructurebased approach and infrastructure-free approach. The authors adopt infrastructure-free approach to develop a real-time route guidance algorithm, called V2R2 (Vehicle-to-Vehicle Real-time Routing). Our simulation results showed that V2R2 algorithm can effectively nd better paths with less travel time than the shortest path. In addition, it can bypass void areas (i.e. the areas containing empty roads) when collecting real-time trafc information. The simulation results validate the efciency and robustness of the proposed V2R2 algorithm.

Introduction

Many densely populated cities around the world suffer from trafc congestion in road systems. It is well known that trafc congestion causes many serious problems, such as fuel consumption, air pollution and the waste of time for vehicle drivers. For example, Schrank and Lomax [1] investigated the trafc congestion problem in 75 US urban areas. The 75 areas used in the analysis include a range of populations from 100 000 to 17 million for the years 1982 to 2001. They found that the total congestion bill for the 75 areas in 2001 came to $69.5 billion, which was the value of 3.5 billion hours of delay and 5.7 billion gallons of excess fuel consumed. To alleviate the problems caused by trafc congestion, two typical approaches are usually employed: widening existing roads or constructing new roads. Unfortunately, both approaches are very expensive solutions [2]. Instead of the two expensive solutions, a much cheaper alternative solution is to use vehicle route guidance systems. Owing to the rapid advances in wireless communications and global position system (GPS)-enabled 870 & The Institution of Engineering and Technology 2010

devices (such as cellular phones, PDAs and car PCs), it is now feasible to make use of route guidance systems at a reasonable price. More importantly, many studies have indicated that route guidance systems can effectively alleviate trafc congestion. For example, Bose and Ioannou showed that automatically guided vehicles can increase the throughput of roadways to double that of manually driven vehicles [3]. Fig. 1 shows the main components of a vehicle route guidance system. In the software part, the route guidance algorithm is used to nd the quickest path for a vehicle from its origin to its destination. The real-time trafc information collector collects real-time trafc information for potential paths between the origin to the destination and then provides this information to the route guidance algorithm. The human computer interface incorporates some multimedia technologies, such as graphical user interface and speech recognition, to provide a user-friendly interface for vehicle drivers. The digital map provides the map information of the road systems. (The digital map IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
real-time trafc information. With this approach, a large number of roadside sensors and communication equipments need to be installed to monitor the trafc condition on each road. The roadside sensors and communication equipments make use of v2r links to collect the required trafc information. On the other hand, the infrastructure-free approach employs vehicle-to-vehicle (v2v) communication. With this approach, there is no need for roadside sensors and communication equipments because all messages are exchanged between vehicles using v2v links. (Section 2 will compare the two types of approaches in greater detail.) Currently, there are several potential technologies for VANET, such as IEEE 802.11a and IEEE 802.11p [4 7]. Obviously, the infrastructurebased approach is a costly solution since it need to install, manage and maintain a large number of roadside sensors. In view of this, we employ v2v communication to develop a real-time vehicle route guidance algorithm, called V2R2 (Vehicle-to-Vehicle Real-time Routing) in this paper. V2R2 consists of two algorithms: V2R2 guidance algorithm and V2R2 detour algorithm. The former is used to nd the quickest path for a guided vehicle. The latter is used to bypass void areas (i.e. the areas containing empty roads without running cars) during the guiding process. The rest of this paper is organised as follows. Section 2 reviews previous related work, Section 3 describes the V2R2 guidance algorithm and Section 4 describes the V2R2 detour algorithm. Section 5 presents the simulation results. Finally, Section 6 concludes this paper.

Figure 1 Main components of a real-time route guidance system can be downloaded from the Internet via online or ofine connections.) In the hardware part, the GPS unit is used to locate the position of the guided vehicle in the road systems, and wireless communication unit is used to transmit control and data messages to other vehicles. Conventional route guidance systems are designed to compute the shortest path to direct a vehicle from its origin to destination. Nevertheless, the shortest path does not necessarily represent the quickest path because of the dynamic trafc conditions on roadways. For this reason, the state-of-the-art route guidance systems make use of vehicular ad hoc network (VANET) to collect real-time trafc information to nd better paths. There are two types of approaches to collecting real-time trafc information: infrastructure-based approach and infrastructure-free approach (see Fig. 2). The infrastructure-based approach employs vehicle-to-roadside (v2r) communication to collect

Related work

The most fundamental distinction of route guidance systems is whether the system uses a static or a dynamic routing algorithm. A static routing algorithm guides a vehicle from its origin to destination without considering the real-time trafc condition of roads. Most static routing algorithms are variants of shortest path algorithms, such as Dijkstras

Figure 2 Vehicle-to-vehicle and vehicle-to-roadside communications


a V2V communication b V2R communication

IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

871

& The Institution of Engineering and Technology 2010

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
algorithm. The main difference between the variants lies in their priority assignment policy for roads. For example, some static routing algorithms may assign a higher priority to a longer but wider road during rush hours. Some static routing algorithms may assign a higher priority to some specic roads according to the preferences of car drivers. In [8, 9], several shortest path algorithms are tested and evaluated on real roadways to nd out which one is more suitable for real-word scenarios. Nevertheless, without realtime trafc information, it is difcult for static routing algorithms to nd good routes for guided vehicles because the trafc conditions of all roads vary with time (e.g. a car accident may cause a road to be congested immediately). To solve this problem, dynamic routing algorithms are developed, which make use of real-time trafc information to nd potentially quicker paths. Several dynamic routing algorithms have been proposed. For example, in [10], the travel time of each road is modelled as a random variable and its realisation can be estimated in advance and made available for vehicle routing. The problem of routing a vehicle in a trafc network is formulated as a dynamic program. The authors apply an approximate probabilistic treatment to the recurrent relations and propose a labelling algorithm to solve the recurrent equations. In [11], a simple trafc ow model based on the V K relationships was constructed, and three types of route choices were compared: shortest distance (SD), shortest time (ST) and shortest time with route information sharing (RIS). Vehicles using RIS give planned route details to the route information server, which then sends accumulated trafc information based on those routes vehicles. Their simulation results indicated that shortest time with RIS routing scheme can improve the trafc efciency. In [12], the authors presented a traveller characteristic-based double hierarchical optimal path algorithm, which can improve the speed of optimal path calculation. Other dynamic routing algorithms can be found in [13 15]. Dynamic route guidance algorithms can be classied into two types of approaches: infrastructure-based and infrastructure-free. Conventional route guidance systems employ infrastructure-based approach. In this approach, a large number of wireless sensors and communication equipments need to be deployed along the roadside to monitor the trafc condition of each road. The sensors and communication equipments make use of v2r communication links to collect the required trafc information. The local trafc information is then transmitted to a centralised trafc information centre (TIC) for further processing. Each vehicle can obtain its required trafc information from TIC via broadcast channels (such as FM radio) or cellular channels (such as GPRS and 3G) in an on-demand fashion. Much research has been devoted to infrastructure-based approach, such as [11, 16, 17]. However, the infrastructure-based approach has several technical drawbacks. First, it is costly to install, manage and maintain a large number of roadside sensors. Second, the trafc information is limited to only the roads with sensor deployment. Third, a TIC usually covers a 872 & The Institution of Engineering and Technology 2010 large area. To avoid the large communication overhead between the TIC and all vehicles, the trafc information is updated periodically, and the updating period is usually on the order of several minutes (e.g. commonly used period include 5, 10 or 15 min). However, such a long updating period cannot provide efcient real-time route guidance for vehicles. Fourth, a disastrous event for TIC (such as re or a network attack) may render the whole trafc information system useless. In view of these drawbacks, many researchers have studied the infrastructure-free approach in the recent past. With infrastructure-free approach, there is no need for roadside sensors, roadside communication equipments and TIC. Vehicles make use of v2v links and multi-hop relay to collect real-time trafc information. Several vehicleto-vehicle-based trafc information systems have been developed. For example, in [18], the authors employ intervehicle communication to develop a SeU-Organising Trafc Information System (SOTIS). In SOTIS, each vehicle monitors the locally observed trafc situation by repeatedly receiving data packets with detailed information from other vehicles. A trafc situation analysis is performed in each vehicle and the result is broadcasted to all vehicles within the transmission range. In [19], the authors discussed the prototype design of a vehicle-to-vehicle-based traveller information system that utilises IEEE 802.11 wireless standards for inter-vehicle communication, and proposed a location-based broadcasting protocol. In [20], the authors proposed a distributed and infrastructure-free mechanism for road density estimation. The proposed method divides a road into multiple cells and employs cell density data packet to collect density information on each road. Vehicles can utilise the trafc density information to avoid congested roads. Much of the literature on infrastructure-free route guidance systems primarily focuses on the construction of vehicle-to-vehicle-based trafc information systems. Little research has been devoted to the overall operation of route guidance, which requires two important building blocks. First, a smart route query and reply strategy is required to nd all potential quicker paths. Second, a detour algorithm is required to bypass void areas that contain empty roads without running cars. Actually, the two parts are the two main contributions of this paper.

V2R2 guidance algorithm

In this section, we detail how the proposed V2R2 guidance algorithm works. In Subsection 3.1, we give an overview of the V2R2 guidance algorithm. In Subsections 3.2 and 3.3, we describe the important techniques that make V2R2 guidance algorithm an efcient routing algorithm.

3.1 Overview of V2R2 guidance algorithm


In this subsection, we give an overview of the operation of the proposed V2R2 guidance algorithm. We assume that IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
each vehicle knows its own geographic position through the GPS unit (see Fig. 1). Also, each vehicle can determine which road segment it is on via the digital map in use (see Fig. 1). In this paper, we dene a road segment as a section of a street between two intersections (see Fig. 4). The V2R2 algorithm consists of two phases: the query phase and reply phase. Initially, the vehicle that needs route guidance, called source node, sends a unicast packet, called route query (RQ) packet, to its destination (see Fig. 3). The RQ packet will travel along the shortest path to the destination. All vehicles in the shortest path will assist relaying the RQ packet using vehicle-to-vehicle communication links. After the RQ packet arrives at the destination region, a special vehicle in that region, called destination node, will calculate the travel time of the shortest path traversed by the RQ packet. By travel time, we mean the estimated time required the guided vehicle to go from the source node to the destination node. After deriving the travel time, the destination node sends a route reply (RR) packet to the source node (see Fig. 3). (The next subsection will explain how to select a vehicle as the destination node.) The RR packet is sent by ooding from the destination node to the source node to nd all potential better routes that take less travel time than the shortest path. More precisely, for each vehicle on potential routes, when it receives RR packet, it will broadcast RR packet to its all nearby vehicles using vehicle-to-vehicle communication links. When RR packets are transmitted and duplicated along the potential routes, RR packets will record the average speed of all traversed road segments. (In fact, the recorded data are the average speed of each cell, which will be explained in the next subsection). Once RR packets encounter void areas (i.e. the areas containing empty roads without running cars), V2R2 detour algorithm will be invoked to help RR packets bypass the void areas. Eventually, when the RR packets arrive at the source node, the source node can easily determine the quickest path to the destination node according to the recorded speed of all traversed road segments on all potential paths. Specically, the quickest path, denoted by p , can be expressed as follows 0 p arg min@
pi [P

where P denotes the set of all potential paths found by RR packet, pi denotes the ith path of set P, ri, j denotes the jth road segment of path pi and T (ri, j ) denotes the travel time required by the guided vehicle to traverse road segment ri, j . However, the ooding of RR packet causes a serious problem since an unlimited ooding scheme will result in signicant communication overhead. In view of this, V2R2 employs a smart ooding strategy that takes advantage of the following techniques: cell-based data aggregation, longroute packet dropping, loop-duplication packet dropping and convergence-duplication packet dropping. These techniques will be explained in the following subsections.

3.2 Cell-based data aggregation and packet delivery


V2R2 employs cell-based trafc information aggregation and delivers RQ and RR packets on a cell-to-cell basis, as is done in [20]. Specically, adjacent vehicles are arranged into a cell because adjacent vehicles usually share the same trafc condition (see Fig. 4). Each road segment is divided into several xed cells. In the digital map, each road is assigned a road ID, each segment of a road is assigned a segment ID and each cell is assigned a cell ID. The triple (road ID, segment ID, cell ID) uniquely identies a cell, and each vehicle can easily nd out its road ID, segment ID and cell ID according to its current geographical position obtained by the GPS unit. In each cell, the vehicle that is the closest to the cell centre will be elected as the cell head. Only cell heads can relay RQ and RR packets; non-head vehicles do not involve in the relay of RQ and RR packets. Each cell head is also responsible for collecting the real-time trafc condition in its cell (i.e. the average driving speed in the cell). The real-time trafc condition is collected by periodic exchange of hello packets broadcasted by all vehicles in the cell; the cell head can then collect the required trafc information from these hello packets. A proper cell size should be chosen such that two consecutive cell heads can directly communicate with each other. Specically, the distance between two consecutive cell heads should not exceed the maximum transmission

X
ri, j [pi

1 T (r i , j )A (1)

Figure 3 Operation overview of V2R2 guidance algorithm IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

Figure 4 Cell-based data aggregation and packet delivery 873

& The Institution of Engineering and Technology 2010

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
Table 1 Packet format of RQ packet Route query packet source vehicle ID (48 bits) source location (64 bits) destination location (64 bits) generating time (32 bits) road ID (32 bits) Table 3 Packet format of hello packet Hello packet driving speed (8 bits) driving direction (3 bits) vehicle ID (48 bits)

packet generating time (32 bits) shortest-path travel time (32 bits) range of the employed wireless communication technology, such as DSRC (dedicated short-range communications) or WiFi. Another important consideration is that if the cell size is too large, it is more difcult to obtain a more accurate estimation of the trafc condition in a cell. On the contrary, if the cell size is too small, then the communication overhead of the cell-to-cell relay increases accordingly. Tables 1 3 show the packet format of RQ , RR and hello packets. The generating time eld denotes the time for generating a packet. Note that all vehicles are automatically synchronised by GPS, which provides accurate time information to all receivers. In practice, the vehicle ID can be represented by the IP or MAC address used by the vehicle without the extra need for vehicle ID addressing. segment ID (8 bits) cell ID (8 bits)

idea behind long-route drop is that there is no need to nd the paths whose total travel time is larger than the travel time of the shortest path because the default path of route guidance systems is the shortest path from the source node to the destination node. With long-route dropping technique, the ooding overhead of RR packet can be largely reduced, which will be validated in Section 5. The other two packet dropping techniques can be explained with reference to Fig. 5. Fig. 5a explains the main idea of loop-duplication drop. In a road graph with loops, the destination node or any intermediate node that forwards RR packet is likely to receive the same RR packet via the loops in the graph. When this happens, the node receiving the loop-back RR packet will drop the packet and stop the ooding process. Fig. 5b explains the main idea of convergence-duplication drop. As shown in Fig. 5b, a node may receive the ooding RR packets from different directions. When this happens, the node will drop all duplicate RR packets that nd the route with longer travel time than that of the RR packets already received. By doing so, we can discard the route with longer travel time and reduce the ooding of RR packets.

3.3 RR packet dropping techniques


To reduce the communication overhead caused by RR packet ooding, V2R2 employs three packet dropping techniques for RR packet: long-route drop, loop-duplication drop and convergence-duplication drop. Recall that, in RQ packet, there is a eld, shortest-path travel time, which records the total travel time along the shortest path from the origin to the destination for the guided vehicle. When the destination node receives the RQ packet, it copies this eld to the generated RR packet. Then, during the ooding process of RR packet, if any intermediate vehicle nds that the accumulated travel time from the destination node to the intermediate node itself has exceeded the total travel time of the shortest path, the vehicle will drop the RR packet and strop the ooding of RR packets. This dropping technique is called long-route drop. The main Table 2 Packet format of RR packet

V2R2 detour algorithm

In the delivery of RQ and RR packets, it is likely that RQ or RR packets encounter void areas (i.e. the areas containing empty roads without running cars). When this happens, the V2R2 detour algorithm will be invoked to bypass the void areas. In Subsection 4.1, we give an overview of V2R2 detour algorithm. In Subsections 4.2 and 4.3, we describe the details of the V2R2 detour algorithm and give some illustrative examples to explain how V2R2 detour algorithm works.

Route reply packet source vehicle ID (48 bits) packet generating time (32 bits) road ID (32 bits) road ID (32 bits) ... segment ID (8 bits) segment ID (8 bits) ... source location (64 bits) shortest-path travel time (32 bits) cell ID (8 bits) cell ID (8 bits) ... avg. speed (8 bits) avg. speed (8 bits) ... destination location (64 bits) accumulated travel time (32 bits) direction (3 bits) direction (3 bits) ...

874 & The Institution of Engineering and Technology 2010

IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org

Figure 5 RR packet duplication problem


a Loop duplication b Convergence duplication

4.1 Overview of V2R2 detour algorithm


The main idea of V2R2 detour algorithm can be explained with reference to Fig. 6. When a vehicle in a road junction forwards RQ or RR packet to other roads in different directions, the empty road will not acknowledge the RQ or RR packet. Consequently, the vehicle in the road junction can nd the existence of an empty road segment. The vehicle discovering the empty road segment is called void discoverer; the junction on which the void discoverer locates is called start point; and the other junction of the rst found empty road segment is called centre point. After the void discover nds the existence of an empty road segment, it starts the V2R2 detour algorithm by sending a right-detour packet (RDP) and a left-detour packet (LDP). The original relayed packet, such as RQ or RR packet, can be encapsulated into the LDP and RDP packet. RDP is used to bypass the void area from the right-hand side (RHS), and LDP is used to bypass the void area from the left-hand side (LHS). (RHS and LHS are viewed from the perspective of centre point.) Initially, RDP packet is forwarded by the void discoverer to its adjacent road junction with the maximum degree, and LDP packet is forwarded to its adjacent road junction with the minimum degree. This is called rst-node (or rst-junction) selection rule for RDP and LDP packet. Since a digital road map can be regarded as a graph, where nodes represent road junctions and edges represent road segments, we use node and junction interchangeably. The meaning of minimum

and maximum degree can be explained with reference to Fig. 7. We dene the line passing through the start point and the centre point as the baseline. The degree, starting from the LHS of the baseline, increases in the anticlockwise direction. RDP and LDP are designed to travel along the void area, and RDP and LDP will converge together to bypass the whole void area. However, if there is no candidate node for either RDP or LDP, a new start point must be selected to ensure the delivery of both RDP and LDP packets. The new start point is selected backward; that is, the previous road junction that immediately follows the current start point in the original packet delivery path will be selected as the new start point. If the new start point also cannot transmit both RDP and LDP packets, the procedure for selecting a new start point will continue in the backward direction until both RDP and LDP packets can be transmitted successfully. In the worst case, the backward selection will trace back to the original node that sends RQ (or RR) packet. When the worse-case happens, the shortest path is simply used as the default quickest path. In practice, the need for multiple backward re-selection rarely happens. In the next subsections, we will explain how RDP and LDP packets are routed along the void areas.

Figure 6 Overview of V2R2 detour algorithm IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

Figure 7 First-node selection for RDP and LDP forwarding 875

& The Institution of Engineering and Technology 2010

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
4.2 RDP routing
RDP routing employs MDRMDL principle for selecting the next road junction for relayed packet. MDRMDL means Minimum-Distance for Right-hand-side candidates and Minimum-Degree for Left-hand-side candidates. This routing principle makes the relayed packet travel along the border of the void area in the clockwise direction. This principle can be explained with reference to Fig. 8. In Fig. 8, the roads in the void area represent empty roads and grey roads represent normal roads with running vehicles. Suppose that point A is the start point. According to the rst-node selection rule, RDP packet is relayed from point A to point B. We dene the line passing through the centre point, point O, and the current position, point B, as the new baseline. The current position is also called current point. The degree, starting from the LHS of the new baseline, increases in the anticlockwise direction. The adjacent junctions of point B, that is, points C, D and E, are classied into RHS candidates and LHS candidates. The candidates whose degree relative to the new baseline is larger than or equal to 1808 are classied as RHS candidates (i.e. points C and D); the candidates whose degree relative to the new baseline is less than 1808 are classied as LHS candidates (i.e. point E). (Again, RHS and LHS are viewed from the perspective of centre point.) According to the MDRMDL rule, RDP routing selects from RHS candidates the point that is the nearest to the centre point. By nearest, we mean the Euclidean distance between the selected point and the centre point is the smallest. In this example, point C will be selected, and RDP packet will be forwarded to road B C. However, if there is no RHS candidate, then RDP routing will select from LHS candidates the point with the minimum degree (i.e. point E in this example). In what follows, we detail how to classify candidates into RHS ones and LHS ones. Again, consider the example shown in Fig. 8. Initially, we calculate the degree of the current point, point B. The degree of point B can be calculated as follows

uB q arctan(jyB yO j=jxB xO j)

3608 2p

(2)

where (xO , yO ) is the 2-D Cartesian coordinate of point O, (xB , yB ) is the coordinate of point B and q 1808. In this example, point B is in the third quadrant. If the current point is in the rst quadrant, then q 08. If the current point is in the second or fourth quadrant, then (2) should be rewritten as follows

uB q arctan(jxB xO j=jyB yO j)

3608 2p

(3)

where q 908 and q 2708 are used for the second and fourth quadrants, respectively. Combining (2) and (3), we obtain a general form for calculating the degree of the current point 8 > > > > > > < > > > > > > : 3608 2p if current point is in quadrant I or III 3608 q arctan(jxB xO j=jyB yO j) 2p if current point is in quadrant II or IV q arctan(jyB yO j=jxB xO j)

uB

(4)

where q is a quadrant-related parameter, which is dened as follows 8 08 > > < 908 q 1808 > > : 2708 if if if if current current current current point point point point is is is is in quadrant I n quadrant II in quadrant III in quadrant IV

(5)

The degree of a forwarding candidate, say point C, can be calculated in a similar way

uC q arctan(jyC yO j=jxC xO j)

3608 2p

(6)

Let the line passing through the centre point, point O, and the current point, point B, be the new baseline, and reset the degree of the new baseline to 08. Then, the new degree of point C relative to the new baseline can be calculated as follows

uC (3608 uc uB )mod 3608

(7)

Because uC (3608 1808 2258)mod 3608 3158, point C is classied into RHS candidates. Similarly, point D is classied into RHS candidates. According to MDRMDL rule, we need to nd out the Euclidean distance from points C and D to the centre point. The Euclidean distance between point C and point O can be calculated as follows dOC q (xO xC )2 (yO yC )2 (8)

Figure 8 Right-detour packet routing 876 & The Institution of Engineering and Technology 2010

IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
Suppose that the length of a road segment is one unit, then dOC 1. Similarly, we can obtain p the distance between point D and point O: dOD 5. If there is no RHS candidate, then point E will be selected since it is the only one LHS candidate in this example. The algorithm of RDP routing is summarised as shown in Fig. 9. Fig. 10 gives a series of examples to explain how the above algorithm works. In Fig. 10, an arbitrary shape of void area is given. Suppose that point S with coordinate (2, 4) is the start point and that point O with coordinate (3, 4) is the centre point. According to the rst-node selection rule of RDP routing (see Fig. 7), RDP packet will be forwarded to point A in (2, 3). For point A in (2, 3), it has two choices for relaying RDP packet: point (1, 3) and point (2, 2). When point A in (2, 3) is the current point, the line O A becomes the new baseline. The degree of point (1, 3) relative to the new baseline is 18.448, and that of point (2, 2) is 341.568. According to the MDRMDL routing rule, point (1, 3) belongs to LHS candidate, and point (2, 2) belongs to RHS candidate. Point B in (2, 2) will be selected as the next node since it belongs to RHS candidate and has the minimum distance. For point B in (2, 2), it has three choices for relaying RDP packet: point (1, 2), point (2, 1) and point (3, 2). When point B in (2, 2) is the current point, the line O B becomes the new baseline. The degree of point (1, 2) relative to the new baseline is 18.448, that of point (2, 1) is 351.878 and that

Figure 10 Examples of RDP and LDP routing of point (3, 2) is 333.448. According to the MDRMDL routing rule, point (1, 2) belongs to LHS candidate, and points (2, 1) and (3, 2) belong to RHS candidates. According to MDRMDL rule, we need to calculate the distance of points (2, 1) and (3, 2) to the centre point. Suppose that the length of a road segment is one unit; for example, the length between points S and O is one. Then, p the distance between point (2, 1) and point O is 10. The distance between point (3, 2) and point O is 2. Point C in (3, 2) will be selected as the next node since it belongs to RHS candidates and has the minimum distance. When this procedure continues, the following selected points are (4, 2), (4, 1), (5, 1) and so on.

4.3 LDP routing


LDP routing employs MDLMDR principle for selecting the next road for relayed packet. MDLMDR means MinimumDistance for Left-hand-side candidates and MaximumDegree for Right-hand-side candidates. This routing principle makes the relayed packet travel along the border of the void area in anticlockwise direction. The MDLMDR principle works in a similar fashion to MDRMDL. There are three differences between RDP routing and LDP routing. First, in RDP routing, RHS candidates are considered rst; instead, in LDP routing, LHS candidates are considered rst. Second, in RDP routing, the candidate point with the smallest distance to the centre node will be selected from RHS candidates; instead, in LDP routing, the candidate point with the largest degree relative to the new base line will be selected from RHS candidates. Third, in RDP routing, the candidate point with the smallest degree relative to the new base line will be selected from LHS candidates; instead, in LDP routing, the candidate point with the smallest distance to the centre node will be selected from LHS candidates. The algorithm of LDP routing is summarised as shown in Fig. 11. To better understand how the above algorithm works, consider again the example shown in Fig. 10. Suppose that point S with coordinate (2, 4) is the start point and that point O with coordinate (3, 4) is the centre point. According to the rst-node selection rule of LDP routing 877

Figure 9 Algorithm of RDP routing IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

& The Institution of Engineering and Technology 2010

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
point O and point (3, 5) is one unit. Point b in (3, 5) will be selected as the next node since it belongs to LHS candidates and has the minimum distance to the centre point. For point b in (3, 5), it has two choices for relaying LDP packet: point (3, 6) and point (4, 5). When point b in (3, 5) becomes the current point, the line O b becomes the new baseline. The degree of point (3, 6) relative to the new baseline is 08, and that of point (4, 5) is 458. According to the MDLMDR routing rule, both points (3, 6) and (4, 5) belong to LHS candidates. The distance between point O and point (3, 6) p is 2 units, and that between point O and point (4, 5) is 2 units. Point c in (4, 5) will be selected as the next node since it belongs to LHS candidates and has the minimum distance to the centre point. When this procedure continues, the following selected points are (4, 6), (5, 6) and so on.

Simulation results

Figure 11 Algorithm of LDP routing (see Fig. 7), LDP packet will be forwarded to point a in (2, 5). For point a in (2, 5), it has three choices for relaying LDP packet: point (1, 5), point (2, 6), and point (3, 5). When point a in (2, 5) is the current point, the line O a becomes the new baseline. The degree of point (1, 5) relative to the new baseline is 341.568, that of point (2, 6) is 18.448 and that of point (3, 5) is 458. According to the MDLMDR routing rule, point (1, 5) belongs to RHS candidate, and points (2, 6) and (3, 5) belong to LHS candidates. Suppose that the distance between two adjacent junctions is one unit; for example, the length between point S and point O is one unit. p Then the distance between point O and point (2, 6) is 5 units, and that between

To demonstrate the performance of the proposed V2R2 guidance algorithm and V2R2 detour algorithm, a series of simulations were conducted. In Subsection 5.1, we describe the simulation environment and examine the efciency of V2R2 guidance algorithm. We compare the driving time of the shortest path and the quickest path. In Subsection 5.2, we examine the capability and efciency of V2R2 detour algorithm to bypass void areas.

5.1 Performance evaluation of V2R2 guidance algorithm


In our simulations, we employ two simulators: Simulation of Urban Mobility (SUMO) [21], an open-source micro-trafc simulator and MObility model generator for VEhicular networks (MOVE) [22], a simulator for generating realistic mobility models for vehicular networks. We generate three gird roadway maps I, II, III with sizes of 5 5 km2, 10 10 km2 and 15 15 km2 (see Figs. 12 14). Each road in the map is set to a two-way road with the maximum speed limit of 100 km/h in both directions. The

Figure 12 Applying V2R2 guidance algorithm to trafc map I (5 5 km2) 878 & The Institution of Engineering and Technology 2010

Figure 13 Applying V2R2 guidance algorithm to trafc map II (10 10 km2) IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
(Owing to limited space, only the shortest and quickest paths are shown in Table 6.) Several observations can be made from the gures and tables. First, V2R2 guidance algorithm can effectively reduce the driving time for the guided vehicle if some light-trafc roads exist. For example, consider the case shown in Fig. 12 and Table 4. Compared to the shortest path, the quickest path found by V2R2 reduces the driving time from 3867 to 1133 s, achieving a reduction of 70% in driving time. Second, not only the quickest path is found by V2R2, but also several potential routes can be found that have less driving time than the shortest path. This useful feature, combined with other factors, like preferred roads, can provide multiple choices for vehicle drivers. Third, although RR packet makes use of ooding, the number of potential routes found by RR packets is small. This validates that the three packet dropping techniques discussed in Section 3.3 can effectively limit the scope of RR packet ooding. Finally, a phenomenon should be noted. The quickest path does not necessarily avoid all heavy-trafc roads. For example, consider the case in Fig. 13, the quickest path does pass a heavy-trafc road from coordinate (15, 7) to (16, 7). The reason is that in order to avoid this heavy-trafc road, a longer route will be selected, thereby counteracting the benets gained by choosing light-trafc roads.

Figure 14 Applying V2R2 guidance algorithm to trafc map III (15 15 km2)

length of a road segment is set to 490 m, and the length of an intersection is set to 10 m. The radius of a cell is set to 500 m. In Figs. 12 and 13, for heavy-trafc road, a vehicle is generated every second; for light-trafc road, a vehicle is generated every ve seconds. In Fig. 14, we reduce the trafc load to half of the original one: for heavy-trafc road, a vehicle is generated every two seconds; for lighttrafc road, a vehicle is generated every 10 s. After setting the trafc conditions, the vehicle and trafc light behaviour are controlled by SUMO. In the rst part of experiments, we investigate the performance of V2R2 guidance algorithm for the three trafc maps shown in Figs. 12 14. Tables 4 6 show all potential routes found by RR packets for the three maps. Table 4 Potential routes found by RR packet for trafc map I Actual driving time (s) 3867 (shortest path) 3464 2859 2772 1385 1137 1135 1133 (quickest path)

5.2 Performance evaluation of V2R2 detour algorithm


In the second part of experiments, we investigate the performance of V2R2 detour algorithm for different shapes of void areas. Three types of shapes are tested: circular shape, rectangular shape and arbitrary shape, as shown in Figs. 15 17. For each type of shape, both RDP routing

Routes found by V2R2 (0,5) ! (1,5) ! (2,5) ! (3,5) ! (4,5) ! (5,5) ! (6,5) ! (7,5) ! (8,5) ! (9,5) ! (10,5) (0,5) ! (1,5) ! (1,6) ! (2,6) ! (3,6) ! (4,6) ! (5,6) ! (6,6) ! (7,6) ! (8,6) ! (9,6) ! (10,6) ! (10,5) (0,5) ! (1,5) ! (1,4) ! (2,4) ! (3,4) ! (4,4) ! (5,4) ! (6,4) ! (7,4) ! (8,4) ! (9,4) ! (10,4) ! (10,5) (0,5) ! (1,5) ! (1,4) ! (2,4) ! (3,4) ! (4,4) ! (5,4) ! (6,4) ! (7,4) ! (8,4) ! (9,4) ! (9,5) ! (10,5) (0,5) ! (1,5) ! (1,4) ! (2,4) ! (3,4) ! (4,4) ! (5,4) ! (5,3) ! (6,3) ! (7,3) ! (8,3) ! (9,3) ! (10,3) ! (10,4) ! (10,5) (0,5) ! (1,5) ! (1,4) ! (2,4) ! (3,4) ! (3,3) ! (4,3) ! (5,3) ! (6,3) ! (7,3) ! (8,3) ! (9,3) ! (10,3) ! (10,4) ! (10,5) (0,5) ! (1,5) ! (1,4) ! (2,4) ! (2,3) ! (3,3) ! (4,3) ! (5,3) ! (6,3) ! (7,3) ! (8,3) ! (9,3) ! (10,3) ! (10,4) ! (10,5) (0,5) ! (1,5) ! (1,4) ! (1,3) ! (2,3) ! (3,3) ! (4,3) ! (5,3) ! (6,3) ! (7,3) ! (8,3) ! (9,3) ! (10,3) ! (10,4) ! (10,5)

IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

879

& The Institution of Engineering and Technology 2010

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
Table 5 Potential routes found by RR packet for the trafc map II Actual driving time(s) 5629 (shortest path) 5480 5254 5252 5165 5290 3300 Routes found by V2R2

(0,10) ! (1,10) ! (2,10) ! (3,10) ! (4,10) ! (5,10) ! (6,10) ! (7,10) ! (8,10) ! (9,10) ! (10,10) ! (11,10) ! (12,10) ! (13,10) ! (14,10) ! (15,10) ! (16,10) ! (17,10) ! (18,10) ! (19,10) ! (20,10) (0,10) ! (1,10) ! (2,10) ! (3,10) ! (3,9) ! (4,9) ! (5,9) ! (6,9) ! (7,9) ! (8,9) ! (9,9) ! (10,9) ! (11,9) ! (12,9) ! (13,9) ! (14,9) ! (15,9) ! (16,9) ! (17,9) ! (18,9) ! (19,9) ! (20,9) ! (20,10) (0,10) ! (0,9) ! (1,9) ! (2,9) ! (3,9) ! (4,9) ! (5,9) ! (6,9) ! (7,9) ! (8,9) ! (9,9) ! (10,9) ! (11,9) ! (12,9) ! (13,9) ! (14,9) ! (15,9) ! (16,9) ! (17,9) ! (18,9) ! (19,9) ! (20,9) ! (20,10) (0,10) ! (1,10) ! (2,10) ! (2,9) ! (3,9) ! (4,9) ! (5,9) ! (6,9) ! (7,9) ! (8,9) ! (9,9) ! (10,9) ! (11,9) ! (12,9) ! (13,9) ! (14,9) ! (15,9) ! (16,9) ! (17,9) ! (18,9) ! (19,9) ! (20,9) ! (20,10) (0,10) ! (1,10) ! (2,10) ! (2,9) ! (3,9) ! (4,9) ! (5,9) ! (6,9) ! (7,9) ! (8,9) ! (9,9) ! (10,9) ! (11,9) ! (12,9) ! (13,9) ! (14,9) ! (15,9) ! (16,9) ! (17,9) ! (18,9) ! (19,9) ! (19,10) ! (20,10) (0,10) ! (0,9) ! (1,9) ! (2,9) ! (3,9) ! (4,9) ! (5,9) ! (6,9) ! (7,9) ! (8,9) ! (9,9) ! (10,9) ! (11,9) ! (12,9) ! (13,9) ! (14,9) ! (15,9) ! (16,9) ! (17,9) ! (18,9) ! (19,9) ! (19,10) ! (20,10) (0,10) ! (0,9) ! (1,9) ! (2,9) ! (3,9) ! (4,9) ! (4,8) ! (5,8) ! (6,8) ! (7,8) ! (8,8) ! (9,8) ! (10,8) ! (11,8) ! (12,8) ! (12,7) ! (13,7) ! (14,7) ! (15,7) ! (16,7) ! (17,7) ! (18,7) ! (19,7) ! (20,7) ! (20,8) ! (20,9) ! (20,10) (0,10) ! (0,9) ! (1,9) ! (2,9) ! (3,9) ! (4,9) ! (4,8) ! (5,8) ! (6,8) ! (7,8) ! (8,8) ! (8,7) ! (9,7) ! (10,7) ! (11,7) ! (12,7) ! (13,7) ! (14,7) ! (15,7) ! (16,7) ! (17,7) ! (18,7) ! (19,7) ! (20,7) ! (20,8) ! (20,9) ! (20,10) (0,10) ! (1,10) ! (2,10) ! (2,9) ! (3,9) ! (4,9) ! (4,8) ! (5,8) ! (6,8) ! (6,7) ! (7,7) ! (8,7) ! (9,7) ! (10,7) ! (11,7) ! (12,7) ! (13,7) ! (14,7) ! (15,7) ! (16,7) ! (17,7) ! (18,7) ! (19,7) ! (20,7) ! (20,8) ! (20,9) ! (20,10) (0,10) ! (0,9) ! (1,9) ! (2,9) ! (3,9) ! (4,9) ! (4,8) ! (5,8) ! (6,8) ! (6,7) ! (7,7) ! (8,7) ! (9,7) ! (10,7) ! (11,7) ! (12,7) ! (13,7) ! (14,7) ! (15,7) ! (16,7) ! (17,7) ! (18,7) ! (19,7) ! (20,7) ! (20,8) ! (20,9) ! (20,10) (0,10) ! (0,9) ! (1,9) ! (2,9) ! (3,9) ! (4,9) ! (4,8) ! (5,8) ! (5,7) ! (6,7) ! (7,7) ! (8,7) ! (9,7) ! (10,7) ! (11,7) ! (12,7) ! (13,7) ! (14,7) ! (15,7) ! (16,7) ! (17,7) ! (18,7) ! (19,7) ! (20,7) ! (20,8) ! (20,9) ! (20,10) (0,10) ! (0,9) ! (1,9) ! (2,9) ! (3,9) ! (4,9) ! (4,8) ! (4,7) ! (5,7) ! (6,7) ! (7,7) ! (8,7) ! (9,7) ! (10,7) ! (11,7) ! (12,7) ! (13,7) ! (14,7) ! (15,7) ! (16,7) ! (17,7) ! (18,7) ! (19,7) ! (20,7) ! (20,8) ! (20,9) ! (20,10) (0,10) ! (0,9) ! (1,9) ! (2,9) ! (3,9) ! (4,9) ! (4,8) ! (5,8) ! (5,7) ! (6,7) ! (7,7) ! (8,7) ! (9,7) ! (10,7) ! (11,7) ! (12,7) ! (13,7) ! (14,7) ! (15,7) ! (16,7) ! (17,7) ! (18,7) ! (18,8) ! (19,8) ! (19,9) ! (19,10) ! (20,10) (0,10) ! (0,9) ! (1,9) ! (2,9) ! (3,9) ! (4,9) ! (4,8) ! (4,7) ! (5,7) ! (6,7) ! (7,7) ! (8,7) ! (9,7) ! (10,7) ! (11,7) ! (12,7) ! (13,7) ! (14,7) ! (15,7) ! (16,7) ! (17,7) ! (18,7) ! (18,8) ! (19,8) ! (19,9) ! (19,10) ! (20,10) area in Fig. 16; and only three non-empty roads is contained in the found void area in Fig. 17. In the third part of experiments, we examined the effect of applying only RDP routing or only LDP routing to nd void areas for the three types of shapes. Owing to limited page space, we only draw the results of applying RDP-only routing and LDP-only routing for circular shape (see IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

3298

2923

2921

2795

2670

2669

2544 (quickest path)

and LDP routing are jointly used to nd void areas. It can be observed from these gures that V2R2 detour algorithm can effectively nd out the void areas and bypass them. Occasionally, the found void area erroneously contains only a few non-empty roads. For example, no non-empty roads are contained in the void area found by LDP and RDP in Fig. 15. Only one non-empty road, which is marked by bold black line, is erroneously contained in the found void 880 & The Institution of Engineering and Technology 2010

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
Table 6 Potential routes found by RR packet for trafc map III (only the shortest path and the quickest path are shown) Actual driving time (s) 3968 (shortest path) 2291 (quickest path) Routes found by V2R2

(0,15) ! (1,15) ! (2,15) ! (3,15) ! (4,15) ! (5,15) ! (6,15) ! (7,15) ! (8,15) ! (9,15) ! (10,15) ! (11,15) ! (12,15) ! (13,15) ! (14,15) ! (15,15) ! (16,15) ! (17,15) ! (18,15) ! (19,15) ! (20,15) ! (21,15) ! (22,15) ! (23,15) ! (24,15) ! (25,15) ! (26,15) ! (27,15) ! (28,15) ! (29,15) ! (30,15) (0,15) ! (1,15) ! (1,14) ! (2,14) ! (2,13) ! (2,12) ! (3,12) ! (4,12) ! (5,12) ! (6,12) ! (7,12) ! (8,12) ! (9,12) ! (10,12) ! (11,12) ! (12,12) ! (13,12) ! (14,12) ! (15,12) ! (15,11) ! (16,11) ! (17,11) ! (18,11) ! (19,11) ! (20,11) ! (21,11) ! (22,11) ! (23,11) ! (24,11) ! (25,11) ! (26,11) ! (26,12) ! (27,12) ! (28,12) ! (29,12) ! (30,12) ! (30,13) ! (30,14) ! (30,15)

Figs. 18 and 19). It can be observed from the Figs. 15, 18 and 19 that the actual number of empty roads in the circular void area is 32. Although both RDP-and-LDP routing and LDPonly routing can nd the accurate number of empty roads (i.e. 32), RDP-only routing erroneously nds one more empty

road, which is marked by bold black line. According to our simulation results, RDP-and-LDP routing outperforms RDP-only routing or LDP-only routing; that is, RDPand-LDP routing has the smallest misjudgment. Table 7 lists the number of empty roads found by RDP-and-LDP routing, RDP-only routing and LDP-only routing for the

Figure 15 Using RDP-and-LDP routing to bypass circularshaped void area

Figure 17 Using RDP-and-LDP routing to bypass arbitraryshaped void area

Figure 16 Using RDP-and-LDP rectangular-shaped void area

routing

to

bypass

Figure 18 Using RDP-only routing to bypass circular-shaped void area 881

IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

& The Institution of Engineering and Technology 2010

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
and MOVE simulators to evaluate the performance of the two algorithms. Our simulation results validate the effectiveness and efciency of both algorithms.

References

[1] SCHRANK D., LOMAX T.: The 2003 annual urban mobility report. Texas Transportation Institute, The Texas A&M University System. http://ntl.bts.gov/lib/24000/24000/ 24010/mobility_report_2003.pdf, September 2003 [2] HEARNE R., SIDDIQUI A.: Issues of dedicated lanes for an automated highway. Proc. of IEEE Conf. on Intelligent Transportation System, November 1997, pp. 619 624 [3] BOSE A. , IOANNOU P.: Evaluation of mixed semi automated manual trafc. Proc. IEEE Int. Conf. Control Applications, September 1998, vol. 2, pp. 868 872 [4] ABDULHAMID H. , ABDEL-RAHEEM E. , TEPE K.E.: Channel estimation for 5.9 GHz dedicated shortrange communications receiver in wireless access vehicular environments, IET Commun., 2007, 1, (6), pp. 1274 1279 [5] ERGEN M., VARAIYA P.: Throughput analysis and admission control for IEEE 802.11a, Mobile Netw. Appl., 2005, 10, (5), pp. 705 716 [6] HARTENSTEIN H., LABERTEAUX K.P. : A tutorial survey on vehicular ad hoc networks, IEEE Commun. Mag., 2008, 46, (6), pp. 164 171 [7] LI F., WANG Y.: Routing in vehicular ad hoc networks: a survey, IEEE Veh. Technol. Mag., 2007, 2, (2), pp. 12 22 [8] ZHAN F.B.: Three fastest shortest path algorithms on real road network: data structure and procedures, J. Geogr. Inf. Dec. Anal., 1997, 1, pp. 69 82 [9] ZHAN F.B. , NOON C.E.: Shortest path algorithms: an evaluation using real road networks, Transp. Sci., 1998, 32, pp. 65 73 [10] FU L.: An adaptive routing algorithm for in-vehicle route guidance systems with real-time information, Transp. Res. Part B: Methodol., 2001, 35, (8), pp. 749765 [11] YAMASHITA T., IZUMI K., KURUMATANI K.: Car navigation with route information sharing for improvement of trafc efciency. Proc. IEEE Seventh Int. Conf. Intelligent Transportation Systems, October 2004, pp. 465470 [12] YANG Z., ZHANG L., WANG J., WANG Y., GUAN Q., FENG J.: Design of intelligent in-vehicle navigation systems for dynamic route guidance with real-time information. Proc. IEEE Int. IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

Figure 19 Using LDP-only routing to bypass circular-shaped void area

Table 7 Number of empty roads found by different detour methods Shape of Number void area of empty roads in void area Number of empty roads found by RDP-andLDP routing 32 32 35 Number of empty roads found by RDP-only routing 33 33 38 Number of empty roads found by LDP-only routing 32 34 35

circle rectangle arbitrary

32 31 32

three types of shapes. These results validate that the proposed V2R2 detour algorithm can effectively nd out the void areas and bypass them with very small misjudgment.

Conclusions

Compared to road construction or reconstruction, real-time vehicle route guidance system is a very cheap and effective approach for alleviating trafc congestion problems in urban areas. There are two types of approaches to real-time vehicle guidance: infrastructure-based and infrastructurefree. Because the former approach is more costly than the latter approach, we employ the latter approach to develop a real-time vehicle route guidance algorithm, called V2R2. The proposed V2R2 algorithm makes use of vehicleto-vehicle communication. It consists of two parts: V2R2 guidance algorithm and V2R2 detour algorithm. The V2R2 guidance algorithm employs a smart and limitedscope ooding strategy to nd out all potential routes that takes less travel time than the default shortest path. The V2R2 detour algorithm employs RDP routing and LDP routing to bypass void areas containing empty roads. We conducted a series of simulation experiments using SUMO 882

& The Institution of Engineering and Technology 2010

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

www.ietdl.org
Conf. Vehicular Electronics and Safety, December 2006, pp. 184 188 [13] CHEN X., YANG Z.S., WANG H.Y.: A multi-agent urban trafc control system cooperated with dynamic route guidance. Proc. Int. Conf. Machine Learning and Cybernetics, 2006, pp. 330 335 [14] KARIMI A., HEGYI A., DE SCHUTTER B., HELLENDOORN J., MIDDELHAM F.: Integrated model predictive control of dynamic route guidance information systems and ramp metering. Proc. IEEE Seventh Int. Conf. Intelligent Transportation Systems, October 2004, pp. 491 496 [15] ZHANG W. , ZHANG Z. , XU J.: Dynamic route guidance using neurodynamic programming. Proc. Sixth Int. Conf. Intelligent Systems Design and Applications, 2006, pp. 1177 1181 [16] SHI X., XU J., XU Y., SONG J.: A simulation study on agentnetwork based route guidance system. Proc. IEEE Conf. Intelligent Transportation Systems, September 2005, pp. 248 253 [17] ZHANG X., HONG J., FAN S., WEI Z., CAO J., REN Y.: A novel realtime trafc information system based on wireless mesh networks. Proc. IEEE Intelligent Transportation Systems Conf., 2007, pp. 618 623 [18] WISCHOFF L., EBNER A., ROHLING H., LOTT M. , HALFMANN R.: SOTIS a self-organizing trafc information system. Proc. IEEE 57th Vehicular Technology Conf., 2003, vol. 4, pp. 2442 2446 [19] ZHANG J., ZILIASKOPOULOS A.K., WEN N., BERRY R.A.: Design and implementation of a vehicle-to-vehicle based trafc information system. Proc. IEEE Conf. Intelligent Transportation Systems, September 2005, pp. 473477 [20] JERBI M., SENOUCI S.M., RASHEED T., GHAMRI-DOUDANE Y.: An infrastructure-free trafc information system for vehicular networks. Proc. IEEE 66th Conf. Vehicular Technology, September 2007, pp. 2086 2090 [21] SUMO simulation of urban mobility. Available at: http://sumo.sourceforge.net/ [22] KARNADI F.K. , MO Z.H., LAN K.C.: Rapid generation of realistic mobility models for VANET. Proc. IEEE Wireless Communications and Networking Conf., March 2007, pp. 2506 2511

IET Commun., 2010, Vol. 4, Iss. 7, pp. 870 883 doi: 10.1049/iet-com.2009.0163

883

& The Institution of Engineering and Technology 2010

Authorized licensed use limited to: College of Engineering - Kottarakkara. Downloaded on July 16,2010 at 10:19:06 UTC from IEEE Xplore. Restrictions apply.

You might also like