ETAR: Efficient Traffic Light Aware Routing Protocol for Vehicular Networks Omar Sami Oubbati∗ , Abderrahmane Lakas†, Nasreddine Lagraa∗ and Mohamed Bachir Yagoubi∗ ∗
†
Laboratory of Computer Science and Mathematics, University of Laghouat, Algeria {s.oubbati, n.lagraa, m.yagoubi}@mail.lagh-univ.dz
College of Information Technology, United Arab Emirates University, PO Box 15551, Al Ain, UAE
[email protected]
Abstract—Routing in Vehicular Ad hoc Networks (VANETs) is an important factor to ensure a reliable and efficient delivery of data packets. In urban environments, routing protocols must efficiently handle the constantly changing network topology and frequent disconnections due to the high mobility and direction changes of vehicles. The challenge is greater when there are traffic lights fixed along intersections which affect directly the mobility and therefore can greatly impact routing in urban areas. In our previous work [1] we have proposed IRTIV (Intelligent Routing protocol using real time Traffic Information in urban Vehicular environment) that takes into account the real time traffic variation without any use of pre-installed infrastructures or additional messages. However, IRTIV does not take into consideration the traffic lights impact. In this paper, we propose ETAR (Efficient Traffic Light Aware Routing Protocol for Vehicular Networks). This protocol’s objective is to find the most stable path for delivering data packets based on traffic lights and traffic density of vehicles using the periodical exchange of Hello messages. We present simulation-based performance results, which show that the proposed protocol increases the packet delivery ratio and reduces the end-to-end delay. Keywords—VANETs, Routing, Urban environment, Real time traffic estimation, Traffic Lights.
I.
A
I NTRODUCTION
P roper
and efficient delivery of data packets to their destination plays a key role in the success of deploying applications on VANETs (Vehicular Ad hoc Networks). In this kind of network, the high mobility of nodes and the limited coverage of the transmission radio cause network fragmentations and frequent topology changes that make routing a very challenging task especially in urban environments where there are many constraints like node distribution, obstacles (buildings), and disrupting factors like traffic lights, etc. Recent developments of routing methods have spawned many proposals for routing protocols for VANETs with a principal objective to solve the problem of the frequent change in the network topology due to the high mobility of the vehicles, and to improve the end-to-end delay [1]–[8]. The majority of the proposed protocols [1]–[6] offer acceptable performance within some situations, but do not fully consider all the aspects of real world situations. One aspect which impacts the dynamics of road traffic and therefore that of the network connectivity is the traffic lights. However, only a few protocols [7], [8] take into consideration the signals of the traffic Lights c 2015 IEEE 978-1-4799-5344-8/15/$31.00
in the routing decision. The problem of disconnection is treated using the traffic density estimation on the segments, which is an important factor that can enhance the performance of delivering packets. The traffic density computing can be achieved using some statistical parameters like in [5], or in a distributed mechanism [4] based on the traffic information exchanged, that will certainly generate a high overhead leading to network congestion. The present paper, presents a novel routing scheme for urban Vehicular Ad hoc Networks called Efficient Traffic light Aware Routing protocols in VANETs (ETAR). This new protocol chooses the most connected segments by taking into account the segment density and the stable segment using the indicated traffic light. This can be achieved, in a real time manner, by exploiting the periodic exchange of Hello messages, and adds to them only a few bytes describing the status on the road. This can reduce the end-to-end delay and the packet losses with no control messages, the protocol is able to find the most connected and the shortest one (By using the Dijkstra Algorithm in terms of the distance from the source to the destination) and tends to avoid a path that can be quickly broken due the high mobility in the presence of other choices. The remainder of this paper is organized as follows. The related works is briefly described in section II. In section III we detail our proposed protocol and the necessary phases to run it. Then, In section IV, the performances of the proposed protocol are evaluated. Finally, we conclude the paper in section V. II.
R ELATED W ORK
Routing in VANET is attracting the attention of many researchers in the last years. One particular class of challenging problems in VANET is routing data in an urban environment. The many routing methods that have been proposed in the literature can be classified into two categories: traffic light aware routing protocols and traffic light unaware protocols. In the following we describe some of the methods proposed. A. Routing Protocols without Traffic Light In this category, two assumptions are made: (i) the use of a location services to discover the destination’s position such as the Grid Location Service (GLS) [10], and (ii) the use of information about the vehicles density, in each road segment, as well as the state of vehicles connectivity. In the later, many techniques have been proposed without taking into account the impact of traffic light on the dynamics of the traffic and
TABLE I: Comparaison of Position-based Routing Protocols in VANETs features and functionalities Multihop Carry & Forward Greedy Forwarding Discovery phase Dijkstra Algorithm Neighbor table Intersection based GPS map Real time traffic Traffic Light impact
GPCR [6] ✓ ✓
Routing Protocols without Traffic Light GSR [9] RBVT [3] GyTAR [4] ✓ ✓ ✓ ✓ ✓ ✓ ✓
VADD [5] ✓ ✓ ✓
Routing Protocols with Traffic Light STAR [7] RLFF [8] ✓ ✓ ✓ ✓ ✓ ✓ ✓
✓ ✓ ✓ ✓
✓ ✓
✓ ✓ ✓ ✓ ✓
✓ ✓ ✓ ✓ ✓
✓ ✓ ✓ ✓
✓ ✓ ✓
✓ ✓
✓
✓
Forwarding) [8] except that this protocol uses a new approach to detect the connectivity on the segment which avoids blind flooding to measure the connectivity by allowing only vehicles having maximum progress from broadcasting source to rebroadcast the connectivity information. However, this technique can suffer from problem of overhead due to the additional message to know the connectivity. Based on the analysis done above, table I summarizes the characteristics of the routing protocols proposed for VANETs.
on the vehicles connectivity such as GyTAR (an improved Greedy Traffic Aware Routing protocol) [4]. This protocol uses a multihop control packet called CDP (Cell Data Packet), which is generated by each vehicle leaving a road segment and reaching a road junction. CDP is used to calculate the number of vehicles between two successive junctions. A vehicle that is planing to send a data packet, first calculates a score-value based on the distance and the traffic density, for every road segment along the path to the final destination. Then, it sends the packet with along the road segment with the highest score. The main drawback of this protocol is the overhead incurred by the broadcast of CDP packets and the collisions that they eventually generate. Another technique called Road-Based using Vehicular Traffic (RBVT) is proposed by [3]. This technique consists of two components: RBVT-R, which uses the reactive broadcasting for discovering and getting the destination location, and RBVTP, which generates periodical connectivity packets that are propagated to the connected path and save the graph which is disseminated to all nodes in the network in order to compute shortest paths to destinations. The source node must initiate a path discovery at each time using the reactive broadcast, and maintain the connected paths with periodical connectivity packets that will certainly cause network congestion.
To give a solution to the protocols drawbacks mentioned above, we developed a new Intersection-based geographical routing protocol that we called ”ETAR” for Efficient Traffic Light Aware Routing Protocol for VANETs. By exchanging modified version of Hello messages, vehicles can determine in distributed manner the real time density and information about connectivity. Therefore, each vehicle, situated on the intersections, can calculate a score using information about segment traffic lights and exchanged between vehicles. This allows to find the shortest and the most reliable path among others available before launching any data delivering.
B. Routing Protocols with Traffic Light
A. Assumptions
The traffic light affects considerably the routing process. In fact, the segments are frequently partitioned due to the impact of traffic light on the mobility of vehicles. Thus, the vehicles tend to cluster in front of two sides in red light segment and the link between vehicles in these two clusters maybe disconnected. However in the green light segment, the connectivity can be broken at any time due to the high mobility. Different techniques are used to handle this problem, like in the STAR (Shortest-Path-Based Traffic-Light-Aware Routing) [7], where a vehicle arriving at the intersection, it checks whether the red light segment which is closer to the destination is connected or not. If it is connected, the packets will be delivered toward this segment. Otherwise, the packets are delivered toward the vehicles on the green light segment that is closer to the destination. The connectivity information is learned through the periodical exchange of Hello messages. As a drawback, STAR does not take into account the real time traffic measurements that can affect considerably the performance of delivering data packets. The similar technique is used in RLFF (Red Light First
For the correct functioning of the protocol, several assumptions are considered :
III.
ETAR: E FFICIENT T RAFFIC L IGHT AWARE ROUTING P ROTOCOL FOR VANET S
•
All the vehicles are equipped with a Global Positioning System (GPS).
•
A video camera is mounted on the dashboard of vehicles inorder to know the traffic light indicated.
•
The knowledge of the destination’s position is ensured thanks to the GLS [10] .
•
All vehicles are equipped with an embedded digital road map to locate the neighboring intersections.
•
A table of neighbors is created and updated periodically by all vehicles.
•
The format of hello message is modified by adding new fields in order to allow vehicles to calculate the total number of vehicles and the knowledge of the connectivity between two successive intersections.
B. ETAR Functionalities To ensure delivering a data packets to their destinations, one of the three following states (cf. Figure 1) is considered to be running: (i) Path Selection : the traffic is estimated continuously by all nodes on the junctions in order to select the appropriate segment to deliver the data packets, (ii) Greedy Forwarding: once the connected segment is selected, the nearest vehicle to the destination is chosen as the next hop forwarding node, this node will repeat the same process until the packet reach the destination. However (iii) Carry and Forward: if there is no connected segment, the vehicle will carry the packet until it reaches the destination or another vehicle nearest to the destination. To ensure fast delivery, we choose the vehicle with highest velocity as carrier of the data packet. Sparse
successive intersections, we have added new fields in the Hello message format which contain the number of the direct neighboring vehicles located in the right (DRN) and the left (DLN), the total number of the vehicles until the next intersection located in the left (VL) and in the right (VR), two flags (Lj and Rj ) to indicate the connectivity between two successive intersections and one bit (TF) to inform all vehicles around of the traffic light shown updated thanks to the video camera mounted on the dashboard of each vehicle. The calculation of the density and the knowledge of the connectivity are done only on the road segment through exchanging of the new modified Hello messages format. Each vehicle broadcasts the connectivity information which indicates whether or not we can reach left or right intersection (Lj and Rj ) and also calculates the total number of vehicles (VL and VR) upon receiving a Hello message from the Farthest neighbor using the following equations 1 :
Intersection Carry & Forward
Path Selection
V L = DLN + V Lf arthest V R = DRN + V Rf arthest
Sparse
-
Destination Area
Sparse
Intersection
Connected path
Connected path
vehicle
(1)
vehicle
Then the neighboring nodes will re-broadcast the Hello message to their neighboring nodes. The process is repeated until the Hello messages are intercepted by the vehicles (X, Z, Y) at the ends of the segment (cf. Figure 4), and they will learn the connectivity information and the number of all vehicles in this segment.
Greedy Forwarding g
Segment
Connected path
R
Fig. 1: States of ETAR.
Z
DLN
DRN VR
VL
Y
1) Real Time Traffic Estimation: The traffic density on the road segment is calculated with a completely distributed manner using Hello messages exchanged periodically among vehicles, The Hello message format is compliant with the ETSI standard [11]. However, some optional fields have not been considered because they are not necessary for our routing protocol (cf. Figure 2). In order to allow vehicles to calculate Hello Message Format Station ID
Message ID
Generation Time
Proto. Version
Reference Position
Frame Body
2 Bytes
2 Bytes
2 Bytes
2 Bytes
8 Bytes
0 – 2312 Bytes
TF 1bit
DLN 1 Byte
Additional Fields DRN VL VR 1 Byte 2 Bytes 2 Bytes
X Driving Direction
Vehicles on int ersect ions
Fart hest vehicle
Broadcaster vehicle
Ordinary vehicle
Int ersect ion
Green Light
Fig. 4: Calculating of the density In another scenario, where the vehicle S (cf. Figure 5) intercepts a Hello message (cf. Figure 6) from the broadcaster vehicle situated at the end of the segment. Segment
Lj 1 bit
Rj 1 bit
DLN
S
TF Meaning 0 Green Light 1 Red Light
Lj 0 0 1 1
Rj Meaning 0 There are no connectivity on the path 1 There are connectivity on the right only 0 There are connectivity on the left only 1 This path is connected
Fig. 2: Hello message format
H
Driving Direct ion
Vehicles on int ersect ions
Fart hest vehicle
Broadcast er vehicle
Ordinary vehicle
Int ersect ion
Fig. 5: Disconnection scenario. their total number and to know the connectivity between two
Red Light
The vehicle S will have a global vision of this segment and immediately deduct that it is partitioned, and the packet cannot reach the other side of the segment (right intersection, Rj = 0) and consequently this segment will be not selected to transit the data packet. TF
DLN
DRN
VL
VR
Lj
Rj
1
1
3
1
3
1
0
Fig. 6: An intercepted hello packet 2) Path Selection: The path selection is carried out only by the (forwarder/source) vehicles located at the intersections because they are the only places where a routing decision is taken (cf. Figure 3). We give a priority to the red light connected segment because there is no mobility and consequently no risk of disconnection. A score is calculated by the (forwarder/source) vehicle at the intersection for each segment by combining several parameters like the traffic density (T otalvehicles ), the connectivity (Lj and Rj ), the shortest (Dijkstra Dw ) and the indicated traffic light (TF). The segment with the highest score is selected for delivering the data packet. The score is calculated as follows : T OT ALV ehicles Scorei = · (Lj × Rj ) · (1 + T F ) Dw Where T OT ALV ehicles = (V L or V R) + 1.
Algorithm 1: ETAR pseudo code C ← The current vehicle; 2 D ← The destination vehicle; 3 J ← The next intersection (Junction); 4 Nc ← The set of one hop neighbors of C; 5 if C = D then 6 Received packet (Success); 7 else 8 if D ∈ Nc then 9 Forward (packet,D); 10 else 11 if Position(C) ∈ Intersection areas then 12 foreach Segmenti do 13 Scorei = T OT ALV ehiclesi · (Lji × Rji ) · (1 + T Fi ) Dwi 1
14 15
16 17 18 19
(2)
20 21
(3)
and Dw is the Dijkstra weight to the destination in term of distance. When the scores are zero for each segment, in this case the (forwarder/source) vehicle will base only on the Dijkstra weight Dw to forward the data packet to the vehicle moving towards the shortest segment to the destination vehicle based on the velocity vector obtained through the updated neighbor table. Pseudo code of our approach is illustrated in Algorithm 1:
J ← M ax of all (Segmenti , Scorei ); // Select the next intersection of the segment with the highest weight if ∃ vehicle ∈ Nc then Greedy Forwarding (packet, vehicle); else Carry and forward (packet,F); Wait For Neighbours(); // Wait for the neighbor moving towards the destination D
IV.
S IMULATION E XPERIMENTS
A. Simulation Environment We carried out the simulation using NS2.34 simulator, and we compared it with AODV [12] and GyTAR [4]. The simulation scenario represents an area 3×3 km2 of the
Source/ Forwarder
Dest inat ion
Sparse
Forwarder
: Ordinary Vehicle
Forwarder
: Dest inat ion Vehicle
: Source/ Forwarder Vehicle
: Greedy Forwarding
Fig. 3: Path Selection.
: Traffic Light
: I nt ersect ion (junct ion)
Value
Simulation area Number of intersection Number of roads Communication range MAC Protocol Frequency Band Number of packets senders Data packet size Number of vehicles Vehicle speed Cycle of traffic light
3000m × 3000m 18 50 300m 802.11 5.15 GHz 35 1 KB 80-250 0-60 Km/h 40 s
B. Simulation Results Figure 7 the PDR is enhanced as vehicles number increases for all the evaluated protocols. When the density of traffic is low, the problem of partition can occur frequently in the network and more packets cannot reach its destinations due to the connectivity in certain parts of segments. However, when the number of vehicles increases the connectivity is improved and the number of lost or dropped packets is reduced significantly. We also can show that ETAR performs better than the other protocols in the delivery ratio in high densities. This is because our approach always finds the closest connected stable path by exploiting the real time traffic density, the Dijkstra weight and the traffic light dynamically at each intersection. The average EED delay of data packets of the three studied protocols is given in figure 8. In general, the comparison shows that ETAR achieves the lowest end-to-end delay for the different vehicles number in reason of the combination between the greedy forwarding, the Dijkstra weight and the indicated traffic light give the shortest stable connected path to the destination vehicle. 1
Packet Delivery Ratio PDR
0,9
0,8 0,7 0,6 0,5
ETAR
0,4
AODV
0,3
GyTAR
0,2
0,1 0 50
100
150
200
250
2
ETAR
1,5
AODV
1
GyTAR
0,5 0 50
100
150
200
250
Number of vehicles
V.
The protocol performance is evaluated based on two parameters (i) Packet Delivery Ratio (PDR): The ratio of delivered data packet to the destination successfully to the total number of packets generated by the source. The greater value of packet delivery ratio means the better performance of the protocol. (ii) End to End Delay (EED) : The average time taken by a data packet to reach the destination successfully.
0
3 2,5
Fig. 8: End-to-End delay vs. vehicle density
TABLE II: Simulation parameters Parameter
3,5
End to End Delay EED (s)
city map. The VanetMobiSim [13] is used as a microscopic mobility simulator to generate the traffic light and mobility Manhattan model according to the map. The road segments are bidirectional, the vehicles speeds are up to 60 km/h, and the number of vehicles is varying between 80 and 250 depending on the adopted scenario. the following table II summarizes the most important parameters:
300
Number of Vehicles
Fig. 7: Packets delivery ratio vs. vehicle density
C ONCLUSION
At each moment, the shortest stable connected path can be determined to forward data packets more efficiently. To this end, vehicles exchange periodically information about traffic density and connectivity. Simulation results shows that ETAR outperforms other protocols in terms of PDR and EED. As future work, we plan to improve ETAR by providing an efficient solution allowing it to deal correctly with the cases of disconnection and sparsity. R EFERENCES [1] O. S. Oubbati, N. Lagraa, A. Lakas, and M. B. Yagoubi, “Irtiv: Intelligent routing protocol using real time traffic information in urban vehicular environment,” in New Technologies, Mobility and Security (NTMS), 2014 6th International Conference on. IEEE, 2014, pp. 1–4. [2] Y. SHI, X.-y. JIN, and S.-z. CHEN, “Agp: an anchor-geography based routing protocol with mobility prediction for vanet in city scenarios,” The Journal of China Universities of Posts and Telecommunications, vol. 18, pp. 112–117, 2011. [3] J. Nzouonta, N. Rajgure, G. Wang, and C. Borcea, “Vanet routing on city roads using real-time vehicular traffic information,” Vehicular Technology, IEEE Transactions on, vol. 58, no. 7, pp. 3609–3626, 2009. [4] M. Jerbi, S.-M. Senouci, T. Rasheed, and Y. Ghamri-Doudane, “Towards efficient geographic routing in urban vehicular networks,” Vehicular Technology, IEEE Transactions on, vol. 58, no. 9, pp. 5048–5059, 2009. [5] J. Zhao and G. Cao, “Vadd: Vehicle-assisted data delivery in vehicular ad hoc networks,” Vehicular Technology, IEEE Transactions on, vol. 57, no. 3, pp. 1910–1922, 2008. [6] C. Lochert, M. Mauve, H. F¨ußler, and H. Hartenstein, “Geographic routing in city scenarios,” ACM SIGMOBILE Mobile Computing and Communications Review, vol. 9, no. 1, pp. 69–72, 2005. [7] J.-J. Chang, Y.-H. Li, W. Liao, and C. Chang, “Intersection-based routing for urban vehicular communications with traffic-light considerations,” Wireless Communications, IEEE, vol. 19, no. 1, pp. 82–88, 2012. [8] L. Ramachandran, S. Sukumaran, and S. R. Sunny, “An intersection based traffic aware routing with low overhead in vanet,” International Journal of Digital Information and Wireless Communications (IJDIWC), vol. 3, no. 2, pp. 50–56, 2013. [9] C. Lochert, H. Hartenstein, J. Tian, H. Fussler, D. Hermann, and M. Mauve, “A routing strategy for vehicular ad hoc networks in city environments,” in Intelligent Vehicles Symposium, 2003. Proceedings. IEEE. IEEE, 2003, pp. 156–161. [10] J. L. J. J. D. SJ, R. K. De Couto David, and R. Morris, “A scalable location service for geographic ad hoc routing,” 2000. [11] T. ETSI, “Intelligent transport systems (its); vehicular communications; basic set of applications; part 2: Specification of cooperative awareness basic service,” Draft ETSI TS, 2011. [12] S. R. Das, E. M. Belding-Royer, and C. E. Perkins, “Ad hoc on-demand distance vector (aodv) routing,” 2003. [13] J. H¨arri, F. Filali, C. Bonnet, and M. Fiore, “Vanetmobisim: generating realistic mobility patterns for vanets,” in Proceedings of the 3rd international workshop on Vehicular ad hoc networks. ACM, 2006, pp. 96–97.