A Novel Opportunistic Routing Protocol Applied to ... - IEEE Xplore

0 downloads 0 Views 221KB Size Report
Lanzhou, China [email protected]. Abstract—Vehicular ad hoc networks (VANETs) applied to specific scenarios are kinds of opportunistic networks.
The 5th International Conference on Computer Science & Education Hefei, China. August 24–27, 2010

ThP6.18

A Novel Opportunistic Routing Protocol Applied to Vehicular Ad Hoc Networks Wang Jian-qiang

Wu Chen-wen

School of Traffic & Transportation Lanzhou Jiaotong University Lanzhou, China [email protected]

School of Electronic & Information Engineering Lanzhou Jiaotong University Lanzhou, China [email protected]

Abstract—Vehicular ad hoc networks (VANETs) applied to specific scenarios are kinds of opportunistic networks. Introducing currently opportunistic routing protocols and actual VANETs characteristics, the paper proposed an opportunistic routing protocol based on the comprehensive utilities of forwarding nodes. The protocol conducted a dimensionless method to deal with these factors including the distance between the forwarding and the destination, the direction arriving the destination, the forwarding velocity vector and traffic conditions. It adopted an eigenvector method to determine each factor weight, and used a weighted sums method to compose each effect factor. In routing process, the node would select the highest utility node to propagate packets. It would use a store-carryforward strategy to transmit packets in spare networks. By means of the NCTUns on the basis of the electronic map of Lanzhou, simulating the working process of the new routing protocol and the testing results indicate that the protocol could apply better to various actual VANETs scenarios.

opportunistic networks, and we could consider introducing opportunistic routing thoughts and the related research results into VANETs. Opportunistic networks do not request fully connected networks. Consequently, it meets the demands of practical networks demands and has important influences on future pervasive computing.

Keywords—vehicular ad hoc networks, routing protocol, opportunistic routing, utility, NCTUns

Currently, there are two kinds of VANETs routing protocols that are position-based routing protocols and opportunistic routing protocols. The latest research trend is to apply electronic map information to realize route. Routing protocols based on network topology, such as AODV and DSR, do not apply to vehicular ad hoc networks [5]. Typical position-based routing protocols mainly contain GPSR (greedy perimeter stateless routing) [6], GSR (geographic source routing) [7], GPCR (greedy perimeter coordinator routing) [8], A-STAR (anchor-based street and traffic aware routing) [9], CAR (connectivity-aware routing) [10] and so on.

I. INTRODUCTION Vehicular ad hoc network (VANET) is an important part of the intelligent transportation systems (ITS), which has huge application potential and commercial value. In VANETs, vehicles are generally equipped with communication equipment that can realize vehicle-to-vehicle communication (V2V) and vehicle-to-roadside communication (V2R) [1]. Realizing vehicle-to-vehicle wireless communication can bring people’s daily life with lots of conveniences. For example, passengers can make use of VANETs to access to Internet and enjoy diversified services such as receiving real-time road conditions, accident warnings, multimedia data download and weather forecast [2]. At the same time, introducing vehicle-tovehicle wireless communication technology has positive effect on collecting vehicle information, aided driving, traffic flow guidance and traffic signals control. VANETs have some unique characteristics, such as rapidly changing of network topology, high speed moving of nodes and the whole network divided into discontinuous parts usually, which must be considered in routing protocol design [3]. In [4], authors pointed that VANET was one kind of uneven density

Present opportunistic routing protocols in VANETs mainly paid attention to disconnectedness of networks. This paper proposed a new routing protocol based on comprehensive utilities of forwarding nodes, which could apply to various actual application scenarios and had universal applicability. According to simulation results of NCTUns simulator on the basis of Lanzhou electronic map, the new protocol had better network performance than others. II. THE RESEARCH PROGRESS OF VANETS OPPERTUNISTIC ROUTING PROTOCOLS

Recent years, some scholars proposed some opportunistic routing protocols which could be applied to VANETs. [11] put forward a routing algorithm based on Motion Vector (MOVE), which was used for specialized sparse VANETs scenarios. The algorithm used the knowledge of neighboring vehicles' velocities and trajectories to predict the optimal forwarding node that would physically travel closest to a fixed message destination. When encountering spare networks, a forwarding node would adopt carry-and-forward routing strategy until a forwarding opportunity was available. In spare networks, the network performance of MOVE algorithm was higher than a greedy, position-based routing algorithm. However, its network performance was common in general conditions and the

Supported by the Natural Science Foundation of Gansu Province in China under Grant No.3ZS062-B25-006; the Science and Technology Planning Project of Lanzhou in China under Grant No.2009-1-5

978-1-4244-6005-2/10/$26.00 ©2010 IEEE

1005

ThP6.18

scenario that the node density was high. [12] proposed a VADD (Vehicle-Assisted Data Delivery) routing protocol which could apply to various VANETs application scenarios and used carry-and-forward strategy in spare networks. In routing process, every node needed to know its geographical location information and the static electronic map furnished with real-time road conditions, and depended on the weights of every road to make choices. Simulation indicated that the protocol had lower transmission delay compared with other routing protocols. VADD was lacking in deployment of city electronic map furnished with road conditions in every vehicle. [13] proposed a SADV (Static-Node Assisted Adaptive Routing Protocol Vehicular Networks) routing protocol. The core thought of this protocol was to place a static assistance forwarding node at every crossing and make use of the node to process packets, which could lower the transmission delay of packets. In routing process, every node needed to know its location information by means of GPS and acquire supports from the static electronic map. This protocol could be applied better to urban scenarios. However, the disadvantages lay in that it practically had difficulty in placing a static node at every crossing, and increased the cost of networks deployment. Meanwhile, the static node also became a bottleneck of the whole networks. [14] proposed a GeOpps (Geographical Opportunistic) routing algorithm. Introducing navigation system into routing process to improve the routing efficiency was its special characteristic. It needed to make use of GPS that could provide accurate location information and the navigation system that could provide the appropriate routing path to arrive the destination. Thus, the next hop could be selected by utility values function. Simulation results indicated that this protocol had better performance in package delivery rates, average number of hops and transmission delay. However, the disadvantages were that the routing process needed the assistance of the navigation system, and it needed to be further perfected in the error recovery mechanism. According to characteristics of available protocols, the new routing protocol proposed by this paper need not use electronic map and other extra facilities. It could adjust to various actual scenarios of VANETs and had characteristics of simpleness, convenience and high-efficiency. III. NEW VANETS OPPORTUNISTIC ROUTING STRATEGY A. Basic Assumption Assumption 1: each vehicle (node) would know its geographical location information that should be real-time updating with the movement of the vehicle. Assumption 2: in routing process, the source vehicle (node) would know the location information of the destination vehicle (node), which should be periodic updating with the movement of the destination.

B. Routing Criterion Analysis In the routing process, there are many practical factors that should be taken into consideration. If we consider only single influencing factor, it will cause some problems, such as local optimization and routing path redundancy, which are the disadvantages in traditional routing protocols, as shown in Figure 1. We will discuss these influencing factors individually.

θ

Figure 1. Typical problems existing in traditional routing 1) The distance between the forwarding node and the destination node In the process of the data propagation, it needs to approach the destination step by step, and the distance is one of the most important measurement indexes. The main factor considered in traditional position-based routing protocols is the distance factor. However, it does not fit absolutely for VANETs. In the propagation process, the distance d i between the forwarding node i ( x i , y i ) and the destination node D ( x D , y D ) can be calculated by (1): (1)

d i = ( xi − x D ) 2 + ( y i − y D ) 2

2) The direction from the forwarding node to the destination node The direction from the forwarding node to the destination node is another important influencing factor. In the propagation process, the direction value can be calculated by the two nodes’ position coordinates. Supposed that the position coordinates of the source node S was ( xS , y S ) and the destination node D was ( x D , y D ) , the direction VSD could be expressed by the

axis angle θ between VSD and X coordinate axis. As in Figure 1, θ can be calculated by (2).

Using GPS is the prevalent method to acquire real-time locating information. Recent years, using some network infrastructures like Wireless Mesh Router, could also realize network nodes real-time locating.

1006

­ ° arctan ° θ = ® ° ° π + arctan ¯

y x

D D

y x

− − D D

y x

S



y x



S

,

( S S

,

x (

D

x



x

S

D



x

> 0)

S

< 0)

(2)

ThP6.18

3) The velocity of forwarding vehicles When selecting a forwarding node, it needs to consider the moving velocity of forwarding nodes. In some cases, the velocity of vehicles (including magnitude and direction) would directly affect the transmission efficiency. As in Figure 2, among the three forwarding nodes a , b , c , source node S should choose b as the next hop node. Though b is the farthest node from the destination, it probably becomes the first node to propagate data to the next hop d .

we can combine all these utilities into a comprehensive utility. Specially, we claim that the value range of each utility is [0, 1]. For the current node S , when depending on distance information to forward data, what is thought the farthest distance is d s which is from node S to the destination node

D , and the nearest distance is zero which represents the destination node D . The distance utility can be gotten from (5). ­d s − di , d s ≥ di ≥ 0 ° U (dis i ) = ® d s °¯ 0, di > d s

(5)

The direction angle from the source node S to the destination node D is θ , and the direction angle from the forwarding node i to the destination node D is θ i . The smaller angle between

coordinates was ( xi ' , yi ' ) , the corresponding time was t ' . Thus, the velocity V i of the node i could be calculated via (3), the angle ϕ i between the velocity vector VDi and X coordinate axis could be calculated via (4).

xi '− xi > 0

θi

­ θi −θ °1 − π , ° U ( ang i ) = ® 2 ° 0, ° ¯

Both the magnitude and direction of vehicle velocity can be calculated via position coordinates of forwarding nodes. Supposed that the new position coordinates of node i was ( xi , yi ) , the corresponding time was t ; the past position

y i '− y i ­ °° arctan x '− x , i i ϕi = ® y '− y °π + arctan i i , xi '− xi ¯°

and

is, the better it is to propagate

packet. When the angle is over 90 e , the propagation is meaningless. The direction utilities can be calculated by (6).

Figure 2. Influencing factor about forwarding nodes velocity

( xi '− xi ) 2 + ( yi '− yi ) 2 Vi = t '−t

θ

(3)

≥ θi −θ ≥ 0

θi −θ >

(6)

π 2

the velocity angle of vehicle i . Using N max to express the neighbor node’s maximum value among all neighbor nodes of the current nodes, the traffic conditions utilities of forwarding nodes can be gotten from (9). U (V i ) =

xi '− xi < 0

­ °1, U (VD i ) = ® °0 , ¯

4) The traffic conditions of forwarding nodes

C. Comprehensive Utility Calculation

2

The velocity magnitude utility of each forwarding node can be gotten from (7), and V max represents vehicle’s maximum velocity magnitude. The velocity direction utility of each forwarding node can be gotten from (8), using ϕ i to express

(4)

In the propagation process, the forwarding node should take traffic flow density into consideration. Forwarding nodes which are in the high traffic flow density area have higher forward success probability than others. The traffic flow density value of the forwarding node can be expressed by itself neighbor nodes number N i . In the practical realizing process, each node makes both N i information and itself position updating-information sent to all its neighbor nodes.

π

Vi V max

ϕi − θ ≤ ϕi − θ >

U (N i) =

(7) π 2

(8)

π

2

Ni N max

(9)

After acquiring every utility, using the weighted sums method can combine each utility into a comprehensive utility UCi . The calculating process is shown in (10), and

U (i ) expresses the above each utility, weight coefficient of each utility. n

UC i = ¦ Wi ⋅ U (i )

1) Synthetic Method

Wi expresses the (10)

i =1

The above discussed factors can affect routing results. In order to confirm the influencing degree of each factor, we adopt dimensionless method to deal with these influencing factors into corresponding utilities in the first place. Secondly,

2) Weight Coefficient Determination Using eigenvector method to calculate the weight of each factor, there are 5 elements needed to be compared with each other.

1007

ThP6.18

(1) Building a judgment matrix Defining aij = Wi / Wj . aij expresses the meaning that the relative importance estimated value about i factor compared to j factor. W i expresses the weight coefficient of i factor.

was 60s, and the source node sent constantly UDP packets to the destination node during simulation stage. The whole testing process was divided into four groups to be conducted, which individually tested the network performance of DSDV, AODV, DSR and the new routing protocol.

Thus, the judgment matrix A consists of the results of these 5 factors multiple comparisons, as in (11). ª a11 «a A = « 21 « # « ¬a 51

a12 a 22 # a 52

" a15 º ªW1 / W1 W1 / W2 " a 25 »» ««W2 / W1 W 2 / W2 = % # » « # # » « " a 55 ¼ ¬W5 / W1 W5 / W2

" W1 / W5 º " W2 / W5 »» % # » » " W5 / W 5 ¼

(11)

After the evaluation via expert estimations, we can get specific values among the judgment matrix A , as in (12). ª 1 «1 / 5 « A = «1 / 9 « «1 / 6 «¬1 / 7

5 1

9 6

6 4

1/ 6 1/4

1 4

1/4 1

1/5

2

1/3

7 º 5 »» 1/ 2» » 3 » 1 »¼

Figure 3. Simulation scenario to test the protocol

(12)

After matrix A solution, we got the maximum characteristic value λ max = 5 .3807 . Then, we substituted the value λmax into C . R . = C . L . / RI . to make the consistency check, obtaining C . R . = 0 . 085 < 1 . It was obvious that the judgment matrix A satisfied the consistency check. Furthermore, we got the characteristic vector ω corresponding to λmax . Due to the vector ω expressed the weight coefficient, it was necessary to normalize the vector ω . The normalization result was ω = ( 0 . 5763 , 0 . 2308 , 0 . 0354 , 0 . 1040 , 0 . 0534 ) T , individually corresponding to each utility U (i ) .

B. Simulation Results and Analysis We selected three network performance indices including the number of dropping packets, the packet collision number and the network throughput, to evaluate the network performance of these four routing protocols. Figure 4 indicated the dropping packets conditions of the four routing protocols in the testing process. It was obvious that the dropping packets number of DSDV protocol was the highest and the dropping packets number of the new protocol was smaller in normal data propagation period, which was lower than AODV protocol and DSR protocol. The Figure showed that the dropping packets number of each protocol was more during the 30s~40s than other periods, because the network was in disconnectedness conditions during that period. There was no link between the sender and the receiver.

IV. SIMULATION EXPERIMENT AND ANALYSIS

1400

This paper adopted the lasted NCTUns 6.0 simulation tool to confirm the new protocol on the basis of Fedora 11. NCTUns, which was developed by Taiwan National Chiao Tung University, was an open source networks simulator applied in transportation field [15]. At present, it has received more and more approval and popularization. Compared with traditional network simulators such as NS-2 etc., NCTUns uses the genuine TCP/IP protocols stack and possesses the advantages of reality, easy configuration and high fidelity [16].

1200

DSDV

A. Simulation Scenario The setting of simulation scenario was based on the actual electronic map of Lanzhou. The simulation region size was 2.8km*1.6km, including the main urban transport artery of Lanzhou. In this region, we deployed 150 virtual vehicles and the range of velocity was from 10mps to 30mps. The moving path relayed on the trajectories of real vehicles, as in Figure 3. The communication protocol applied in vehicles was IEEE802.11b. In the simulation process, considering that obstacles had effect on communications thus the signals could only transmit along the road directions. The simulation time

AODV DSR

Drop packets

1000

New

800 600 400 200 0 0

20

40

60

Time(s)

Figure 4. The dropping packets conditions Figure 5 illustrated the packets collision conditions of four routing protocols in MAC layer. Generally speaking, if the routing protocol adopted broadcast way to send packets, there were more collision times in MAC layer. Due to what the new protocol adopted was the position-based forwarding mechanism, the packets in MAC layer had less collision times, which had optimal performance in the four testing protocols.

1008

ThP6.18

traditional routing protocols, the new protocol had better network performance and could apply better to VANETs.

400 350

Collision packets

300

DSDV AODV DSR New

REFERENCES [1]

250 200

[2]

150 100

[3]

50 0 0

20

40

60

[4]

Time(s)

Figure 5. Packets collision conditions

[5]

Figure 6 illustrated the receiver throughput conditions of the four protocols in testing process. From the figure, we could see that the throughput of the new protocol was the highest, meaning that the receiver could receive more packets successfully in the unit time. The distance between the sender and the receiver affected the throughput seriously: the closer the distance was, the higher the throughput was. Besides, the throughput was also affected by vehicle velocity and network topological changes. From the practical simulation results, the network performance of the new protocol had advantages over the other three traditional routing protocols, which could be applied better to VANETs and the individual performance index could up to the application demands.

[6]

[7]

[8]

[9]

[10]

450 400

Throughput(KB/s)

350 300

[11]

DSDV AODV DSR New

[12] 250 200

[13] 150 100 50 0 0

[14] 20

40

60

Time(s)

[15]

Figure 6. The throughput of the receiver

[16]

V. ENDING WORDS Importing the multi-attribute utility theory and the opportunistic routing mechanism into the VANETs routing process had efficient effect on improving the efficiency and reliability of the data propagation. This paper preliminarily explored the completely new binding thought and presented the binding routing model and working mechanism. The simulation results indicated that compared with some

1009

Azzedine Boukerche, Horacio A.B.F. Oliveira, Eduardo F. Nakamura, and Antonio A.F. Loureiro, “Vehicular ad hoc networks: a new challenge for localization-based systems,” in Computer Communications, vol. 31, 2008, pp. 2838̄2849. Carlos J. Bernardos, Ignacio Soto, Maria Calderon, Fernando Boavida, and Arturo Azcorra, “VARON: vehicular ad hoc route optimisation for NEMO,” in Computer Communications, vol. 30, 2007, pp.1765̄1784. Y.S. Chen, Y.W. Lin, S.L. Lee, “A mobicast routing protocol in vehicular Ad-Hoc networks,” in Mobile Networks and Applications,vol. 15, no. 1, 2010, pp. 20-35. Mahmoud Abuelela, Stephan Olariu, and Ivan Stojmenovic, “OPERA: Opportunistic packet relaying in disconnected vehicular ad hoc networks,” in IEEE MASS ’08, 2008, pp. 285–294. Fan Li, Yu Wang, “Routing in vehicular ad hoc networks: a survey,” in IEEE Vehicular Technology Magazine, vol. 6, 2007, pp. 12-22. B. Karp, 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, ACM, New York, NY, USA, 2000. C. Lochert, H. Hartenstein, J. Tian, H. Füÿler, D. Hermann and M. Mauve, “A routing strategy for vehicular ad hoc networks in city environments,” in Proceedings of the IEEE Intelligent Vehicles Symposium, 2003. C. Lochert, M. Mauve, H. Füÿler, H. Hartenstein, “Geographic routing in city scenarios,” in ACM SIGMOBILE Mobile Computing and Communications Review , vol. 9, no. 1, 2005, pp. 69-72. G. Liu, B.S. Lee, B.C. Seet, C.H. Foh, K.J. Wong, and K.K. Lee, “A routing strategy for metropolis vehicular communications,” in International Conference on Information Networking (ICOIN), 2004, pp. 134-143. V. Naumov, T. Gross, “Connectivity-aware routing (CAR) in vehicular ad-hoc networks,” in Proceedings of the 26th IEEE International Conference on Computer Communications, 2007, pp. 1919-1927. J. LeBrun, C.N. Chuah, D. Ghosal, M. Zhang, “Knowledge-based opportunistic forwarding in vehicular wireless ad hoc networks,” in Proceedings of the 61st IEEE Vehicular Technology Conference, vol. 4, 2005. J. Zhao, G. Cao, “VADD: Vehicle-assisted data delivery in vehicular ad hoc networks,” in IEEE transactions on vehicular technology, vol. 57, no. 3, 2008, pp. 1910-1922. Y. Ding, C. Wang, L. Xiao, “A static-node assisted adaptive routing protocol in vehicular networks,” in Proceedings of The Fourth ACM International Workshop On Vehicular Ad Hoc Networks, VANET'07, ACM, New York, NY, USA, 2007. I. Leontiadis, C. Mascolo, “GeOpps: Geographical opportunistic routing for vehicular networks,” in Proceedings of the IEEE International Symposium on World of Wireless, Mobile and Multimedia Networks, 2007. S.Y. Wang, C.L. Chou, C.H. Huang, et al. The design and implementation of the NCTUns 1.0 network simulator. Computer Networks, 2003, 42: 175-197. S.Y. Wang, C.L. Chou. NCTUns Tool for Wireless Vehicular Communication Network Researches. Simulation Modeling Practice and Theory, 2009, 31(2): 105-116.