Wireless Personal Communications

, Volume 100, Issue 4, pp 1339–1353 | Cite as

A Connectivity-Based Multi-Lane Routing Optimization Algorithm in Vehicular Communication

  • Hai-tao Zhao
  • Huimin Wang
  • Hongbo Zhu
  • Dapeng Li


This paper proposed a connectivity-based multi-lane geographic routing protocol (CGRP) for vehicular ad hoc networks. The proposed CGRP is based on an effective selection of road intersections through which a package must pass from source to destination. The cooperative connectivity probability and delay are taken into consideration when choosing the most suitable path for delay-sensitive safety traffic. Analytical expressions for cooperative connectivity probability is derived based on a three-lanes path model. Geographical forwarding is used to transfer packets between any two intersections on the path, reducing the path sensitivity to individual node movements. Furthermore, forwarding packets between two adjacent intersections also depend on geographic location information. Neighbor nodes’ priority are assigned according to position, speed, direction and other factors. Node with the highest priority will be selected as the next hop. Numerical and simulation results show that the proposed algorithm outperforms the exsiting routing protocols in terms of the end-to-end delay and the number of hops with a little cost of routing overhead in city environments.


Vehicular ad-hoc network GRP Connectivity probability Priority 

1 Introduction and Related Works

Much existing research considers vehicular ad hoc networks (VANETs) as a vehicle-to-vehicle or a vehicle-to-roadside-unit network architecture that can be easily deployed without relying on expensive network infrastructure. Nevertheless, the communication between vehicles and preexisting fixed infrastructure such as gateways to the Internet opens plethora of interesting applications to both drivers and passengers. The promising applications and the cost effectiveness of VANETs constitute major motivations behind increasing interest in such networks. The success of VANETs revolves around a number of key elements such as the routing message between mobile nodes (MNs) and the gateway to the Internet. Without an effective routing strategy, the success of VANETs will be limited.

There are more issues, such as unpredictable drivers behavior [1, 2], should be considered in the design of routing protocols. The existing routing protocols are usually designed with the assumption that the vehicles are uniformly random distributed on the road [3] which is inconsistent with the actual case. Hence, routing protocols that simply consider the average vehicle density or probability of connectivity may choose the improper road which results in disconnection problems. In order to solve the aforementioned problems, a well-designed routing protocol often consists of two steps: (1) select an optimal route, consisting of a sequence of passed road intersections; (2) select the next hop, usually by the way of greedy forwarding.

In [4, 5],the poor performance of the traditional routing protocols for mobile ad-hoc networks in VANET was demonstrated. In [6, 7, 8], some topology-based routing protocols such as optimized link-state routing (OLSR), dynamic source routing and ad-hoc on demand distance vector routing (AODV) in VANET environments were proposed, the main problem with these protocols is their route instability. Indeed, the traditional node-centric view of the routes (i.e., an established route is a fixed succession of nodes between the source and destination) leads to frequent broken routes in the presence of VANET high mobility. Consequently, many packets are dropped, and the overhead due to route repairs or failure notifications significantly increases, leading to low delivery ratios and high transmission delays [9]. Location-based routing protocol can be a good solution to this problem. The data transmission from source node to destination node only need to know location of the destination node and the geographical location of the next-hop node without any other topology information [10, 11, 12] which is more adaptive for the VANET. Some typical and popular routing schemes based on position in VANET are discussed. In [13], the junction-based geographic routing (JBR) which is one of the latest published junction-based routing is proposed makes use of selective greedy forwarding up to the node that is located at a junction and is closer to the destination. In [14],a junction-based multipath source routing algorithm was proposed to alleviate the issue of local optimum. In [15], the impact of traffic light on routing protocol design was investigated based on an intersection based routing protocol designed for vehicular communications in urban areas. In [16], Greedy perimeter stateless routing (GPSR) was proposed as a typical position-based routing which uses greedy forwarding to forward packets initially. When a packet reaches a local optimum, it switches to the perimeter mode. Connectivity are not taken into account in these protocols.

Differing from the exiting routing protocols, A connectivity-based multi-lane geographic routing protocol (CGRP) for vehicular ad hoc networks was proposed. The proposed CGRP selects the lanes with highest connectivity probability to forward data. In addition, a three-lanes path model is proposed. The expression of connectivity probability of assistance in each lane is calculated based on this model. The mechanism of the next jump is also improved.

The remainder of the paper is organized as follows: Sect. 2 and 3 describes CGRP including the chosen of lane in routing discover and the chosen of next dance. Simulation results are presented in Sect. 4 with some analysis following the conclusion and cited references.

2 System Model of Three-Lane

A typical VANET environment which consists of roads with intersections is proposed. The street map is abstracted as a graph G(V, E). For any two intersections A and B, (A, B)∈G, only when there is a road segment connecting A and B that vehicles can travel on that segment. A model of a three-lane road segments was proposed in Fig. 1.
Fig. 1

Three-lane road segments structure

Assuming that each vehicle in certain lane moves in a constant speed and Lane 1, 2, 3 correspond to the speeds of V1, V2, V3 respectively. Supposing that the number of nodes in three lanes all obey the Poisson distribution [17] with the density of K1, K2, K3. Also, paths are divided into several segments and the length of the segments is half of the communication range. The nodes of the segments is independent and identically distributed, obeying the Poisson distribution with parameters of (R/2) * K i . The probability distribution of the number of nodes in each segment is as follows:
$$P\left( {N_{i} = n_{i} } \right) = \frac{{\left( {\left( {{\raise0.7ex\hbox{$R$} \!\mathord{\left/ {\vphantom {R 2}}\right.\kern-0pt} \!\lower0.7ex\hbox{$2$}}} \right)*Ki} \right)^{{n_{i} }} }}{{n_{i} !}}e^{{ - \left( {{\raise0.7ex\hbox{$R$} \!\mathord{\left/ {\vphantom {R 2}}\right.\kern-0pt} \!\lower0.7ex\hbox{$2$}}} \right)*k_{i} }}$$
Considering the scenario shown in Fig. 1, vehicle 1 cannot communicate with vehicle 2 directly because the distance between them is greater than communication range R. However they can communicate indirectly with the help of vehicle 3 and vehicle 4. View the three lanes as a lane, as long as each segment has a vehicle, the connectivity of the network can be guaranteed (lateral distance has been ignored). The connectivity probability between vehicles 1 and 2 can be presented as follows:
$$P_{C} = \left( {1 - P\left( {N_{3} = 0} \right)} \right)^{2} = e^{{ - R*K_{3} }}$$
In this formula, \(N_{3}\) represents the number of vehicles in certain segment of lane 3. R is the communication range. \(K_{3}\) is the density of vehicles in lane 3. In general case (here we only consider the vehicles indirectly communicate in lane 1 with the aid of vehicles in lane 2 and lane 3). In the case of that there are two segments between the disconnected adjacent vehicles on the lane 1 (n = 2), vehicles in lane 1 can communicate indirectly with the aids with lane 2 and lane 3. Therefore, when n = 2, the connectivity probability P C can be expressed as formula (3).
$$P_{C} = \mathop C\nolimits_{2}^{2} \left[ {1 - f\left( {N_{2} = 0} \right)} \right]^{2} + \mathop C\nolimits_{2}^{1} \left[ {1 - f\left( {N_{2} = 0} \right)} \right]\left[ {1 - f\left( {N_{3} = 0} \right)} \right] + \mathop C\nolimits_{2}^{0} \left[ {1 - f\left( {N_{3} = 0} \right)} \right]^{2}$$
By that analogy, when n = 3, the connectivity probability P C can be expressed as formula (4).
$$\begin{aligned} P_{C} & = \mathop C\nolimits_{3}^{3} \left[ {1 - f\left( {N_{2} = 0} \right)} \right]^{3} + \mathop C\nolimits_{3}^{2} \left[ {1 - f\left( {N_{2} = 0} \right)} \right]^{2} \left[ {1 - f\left( {N_{3} = 0} \right)} \right] \\ &\quad + \,\mathop C\nolimits_{3}^{1} \left[ {1 - f\left( {N_{2} = 0} \right)} \right]\left[ {1 - f\left( {N_{3} = 0} \right)} \right]^{2} + \mathop C\nolimits_{3}^{0} \left[ {1 - f\left( {N_{3} = 0} \right)} \right]^{3} \\ \end{aligned}$$
when n = k, the connectivity probability P C can be expressed as formula (5).
$$P_{C} = \sum\limits_{i = 0}^{k} {\mathop C\nolimits_{k}^{i} } \left[ {1 - f\left( {N_{2} = 0} \right)} \right]^{i} \left[ {1 - f\left( {N_{3} = 0} \right)} \right]^{K - i}$$
If X i represents the distance between adjoining nodes in lane 1, the number of segments between them is 2X i /R. The formula (5) can be presented as follows.
$$P_{C} = \sum\limits_{i = 0}^{{X_{i} /(R/2)}} {C_{{X_{i} /R/2}}^{i} } \left[ {1 - f\left( {N_{2} = 0} \right)} \right]^{i} \left[ {1 - f\left( {N_{3} = 0} \right)} \right]^{{X_{i} /(R/2) - i}}$$
The distance X1 between vehicles in lane 1 is subject to the exponential distribution with parameters of vehicle density K1.
$$F_{{X_{1} }} \left( x \right) = 1{ - }e^{{ - K_{1} *x}}$$
Assuming that K1 is 8 vehicles/km and drew the connectivity probability distribution with the distance between the adjacent vehicles as Fig. 2.
Fig. 2

Probability distribution of distance between vehicles

It can be found from Fig. 2 that the possibility is approximately up to 1 when the distance is greater than 400 meters. X1 subjects to the exponential distribution with parameters of K1, So the interrupt probability of lane 1 is as bellows:
$$P_{d} = P\left( {X_{1} > R} \right) = e^{{ - R*K_{1} }}$$
Considering that there may have multiple disconnected links in lane 1. The number of disconnected link and the number of nodes are denoted as q and Q respectively, \((q \in 1,2, \ldots Q - 1)\). The probability that there are q links been assisted connected is accumulation.
The probability that there are q links are disconnected in lane 1 without considering the collaboration between the lanes is shown as formula (9).
$$P_{q} = C_{Q - 1}^{\text{q}} P_{d}^{q} \left( {1 - P_{d} } \right)^{Q - 1 - q}$$
So the connectivity probability of lane 1 is presented as formula (10).
$$P_{1C} = \sum\limits_{q = 0}^{Q - 1} {P_{cq} *P_{q} }$$
Similarly, the connectivity probability of lane 2 and lane 3 can be computed as lane 1 between adjacent gateways and be denoted as \(P_{2C}\) and \(P_{3C}\). The gateway will select the lane which has the highest assisted connectivity probability to forward packages.
$$P_{s} = Max\{ P_{1C} P_{2C} P_{3C} \}$$
As is known that the path is composed of several successive segments, the probability of a path is eventually calculated as formula (12)
$$P = \prod\limits_{{{\text{j}} = 1}}^{\text{n}} {P_{\text{s}} }$$

3 Connectivity-Based Multi-Lane Routing Optimization Algorithm

The proposed CGRP is introduced in this section. Firstly, the system model is proposed to build the framework. Then, the functionality of CGRP is presented. It is assumed that the location-aware vehicles which obtain their geographical position from a global positioning system (GPS) receiver or other location service such as in [18]. Each node maintains neighbor table which includes location information of neighbor nodes. In the scenario, the network consists of MNs (vehicles) including source and destination (see Fig. 3). When a message is generated at an MN, depending on its location, it will be relayed multiple times through several vehicles before reaching the destination.
Fig. 3

The message transmitting procedure in VANETs using CGRP

3.1 Backbone Routing Discover

Supposing that every crossroad was equipped with a gateway. Source node will send out a routing discover packet. Each neighbor node which received the discover packet will rebroadcast it, unless it is the gateway or node which has a route to the destination in its route cache. The gateway receives the discover packet which contains the position of destination and communicate with other gateways about the connectivity probability of the route and the corresponding time delay. Then, it sends a route reply packet back to the source node. Upon the arrival of the route reply packet at the source node, the source node begins to send data along the selected path. The calculation of time delay is different due to timestamps.

3.2 The Choice of the Next Dance

As nodes are rapidly moving vehicles, in the actual forwarding process, in order to guarantee the connectivity of network and the QoS indicators, a CGRP forwarding mechanism is proposed in this section, that is, the choice of next hop is considered in this section. According to the specific conditions of roads in urban areas, these two scenarios will be discussed respectively: the straight roads and the crossroads.

3.2.1 Selection of Next Dance on Straight Road

The performance of routing protocols based on location information applied to high-speed mobile nodes is not particularly satisfactory in the VANET. Due to the high-speed movement of nodes, the information of routing table can not accurately reflect the location information of the neighbor node. And due to the limited wireless resources, the update period of neighbor table can not be set too small. Also, the communication distance of sending nodes in a straight-ahead road is 200 m. In the routing protocol, the period for sending beacon frames is 1 s, that is, the location information in the neighbor table of the sending node is updated every second, assuming that the update time of the last beacon frame is t, and when the sending node sends data at time t′, t′ − t = 0.5 < 1 s, the speed of the neighboring node is 20 m/s, if the coordinate position in the neighbor table is 195 m away from the sending node, The distance between the sending node and the node is 205 m, which is no longer within the communication range of the sending node. At this time, the node following the GRP will roll back the data packet to the previous node according to the record of the back-off table, and the pervious node will reselect the next hop, if failed, the data packet will fall back to the previous node for the next hop reselection, so that the data packet will not be sent until the data is rolled back to the source node. The situation is even more obvious if the condition that the sending node is traveling in the opposite direction to the node is considered. In order to reduce the number of forwarding hops and delay and increase the success rate of data transmission, a strategy of assigning priority to neighbor nodes on the straight road which takes the driving direction, position and speed of the neighbor nodes into account is proposed.

Since the neighbor table only shows the time and location information without the speed and direction information, it is necessary to calculate the distance between each neighbor node and the sending node and the speed of neighbor node. By adding the information of these two fields to the neighbor table, the values of these three fields at the neighbor table update time can be calculated from Table 1:
Table 1

Improved neighbor table on straight road






IP address and type









Updated time



Expiration Time



Distance between source nodes and neighbor nodes



Speed of nodes

Upon obtaining its own position(self_lat, self_long) through GPS and the latitude and longitude position (nbri_lat, nbri_long) of neighbor nodes from the Neighbor Table, the source node can calculate the distance from each neighbor node to itself as follows:
$$L_{\text{i}} = \sqrt {\left( {{\text{self}}\_lat - nbri\_lat} \right)^{2} + \left( {{\text{self}}\_long - nbri\_long} \right)^{{^{2} }} }$$
In order to reduce the number of hops and try to forward the data packets to the nodes far away from them, the higher priority is assigned to the nodes far away, and the distance factor of the priority \(a_{i}\) is as follows:
$$a_{i} = {\raise0.7ex\hbox{${\text{R}}$} \!\mathord{\left/ {\vphantom {{\text{R}} {L_{i} }}}\right.\kern-0pt} \!\lower0.7ex\hbox{${L_{i} }$}}$$
High mobility can increase network load, which can lead to link failures. So the node with the highest priority is assigned with the slowest speed. The sending node can obtain the latitude and longitude of the neighbor node from the neighbor table at timestamp 1 and timestamp 2 (assuming uniform motion) respectively. The speed of neighbor i is as follows:
$$V_{i} = \frac{{\sqrt {\left( {nbri\_lat2 - nbri\_lat1} \right)^{2} + \left( {nbri\_long2 - nbri\_long1} \right)^{2} } }}{TS2 - TS1}$$
where (nbri_lat1, nbri_long1) and (nbri_lat2, nbri_long2) represents the latitude and longitude of neighbor nodes from the neighbor table at timestamp 1 and timestamp 2.
Assuming that \(V_{{M{\text{ax}}}}\) is the limited speed on the selected lane, the factor of the priority speed \(b_{i}\) is as follows:
$$b_{i} = {\raise0.7ex\hbox{${V_{Max} }$} \!\mathord{\left/ {\vphantom {{V_{Max} } {V_{i} }}}\right.\kern-0pt} \!\lower0.7ex\hbox{${V_{i} }$}}$$
Therefore, the priority of the vehicle node on the straight road can be set by the following formula:
$$pri = a_{i} \times b_{i}$$

3.2.2 Selection of Next Dance on Crossroads

The choice of next dance on crossroads is more complex than that on the straight roads for the uncertain directions in the next moment. The information field of angle is increased in the neighbor table as follows (Table 2).
Table 2

Improved neighbor table on crossroads


Angle (\(\theta\))

The angle between the L1 and L2 (assume that the connection between the source to the destination node is L1 and the connection between the source to the neighbor node is L2)

Supposing that \(\left( {nbri\_lat1,nbri\_long1} \right)\) and \(\left( {nbri\_lat2,nbri\_long2} \right)\) are the positions of ith neighbor node at timestamp1 and timestamp2. Then \(\theta_{i}\) can be calculated based on them by formula (18):
$$\theta_{i} = \arctan \frac{nbri\_long2 - nbri\_long1}{nbri\_lat2 - nbri\_lat1}$$
And source node can obtain position (self_lat1, self_long1) of itself by GPS. So the angle between L1 and longitude can be calculated by formula (19).
$$\theta_{SD} = {\text{arctant}}\frac{des\_long1 - self\_long1}{{{\text{des\_lat1 - self\_lat1}}}}$$

Finally, \(\theta\) can be presented as \(\theta_{i} - \theta_{SD}\).

So the priority factor \(C_{\text{i}}\) is calculated by formula (20):
$${\text{c}}_{i} = {\raise0.7ex\hbox{$c$} \!\mathord{\left/ {\vphantom {c {\left( {\theta_{i} - \theta_{SD} } \right)}}}\right.\kern-0pt} \!\lower0.7ex\hbox{${\left( {\theta_{i} - \theta_{SD} } \right)}$}}$$
Where c is a constant. In summary, the following the priorities of the neighbor nodes at the crossroads are assigned as follows:
$$pri = a_{i} *b_{i} *c_{i}$$
The flow chart of proposed CGRP is shown as follows:

4 Simulation and Analysis

The proposed protocol is implemented on OPNET and VanetMobiSim which generates moving trace. The protocol performance is evaluated by package delivery rate, average transmission delay and routing overhead. It can be concluded that the distance between the two adjacent vehicles on lane 1 is greater than 400 m approximately from Fig. 2. So 400 and 600 m is putted in this part and other road parameters are shown in Table 3.
Table 3

Simulation parameters

Simulation parameters


Vehicles density on lane 1: K1

8 vehicles/km

Distance of adjacent nodes on lane 1: X1

400, 600 m

Number of segments between adjacent nodes: n

2, 3

Communication range: R

400 m

Vehicles density on lane 2: K2

5 vehicles/km

Vehicles density on lane 3: K3


Connectivity probability on lane 1: P C

The dependent variable

The following Fig. 4 shows the relationship between the density of the vehicles on lane 3 (K3) and connectivity probability on lane 1 (P1C). The curve expresses that there are two segments between two adjacent vehicles on lane 1, while the following curve represents three segments. The figure shows that the bigger K3 is, the higher the assisted connectivity probability lane 1 gets under the situation of unchanged K2. With the increment of the number of the vehicles on lane 3, the probability of disconnected vehicles being able to connect indirectly via vehicles on lane 2 and 3 become higher. It is also concluded that with the growth of the distance between the adjacent inter-vehicle, the collaborative communication becomes more difficult to realize.
Fig. 4

The relationship between the density of the vehicles on lane 3 and connectivity probability on lane 1

Figure 5 was drawn with K3 on the horizontal axis and communication range R on the vertical one. The desired R decreases with the increasing of K3. On the other hand, the bigger the R is, the greater the probability of communication on lane 1 will be without the change of K3. As shown in Fig. 6, connectivity probability on lane 1 increases as the communication range R increases. Connectivity probability is close to 1 when the value of K3 is 0.008 and value of R is 200 m while the value of K3 is 0.006 and value of R is 400 m. To meet the connectivity probability of lane 1, gateway can determine the communication range of vehicles dynamically according to vehicle density on lane 3.
Fig. 5

The relationship between the density of the vehicles on lane 3 and the communication range

Fig. 6

The relationship between the communication range and the connectivity probability on lane 1

The proposed protocol is implemented on OPNET and VanetMobiSim which generates moving trace. Our improved protocol is compared with the other three protocols, GPSR, GPCR, and OLSR. We use the international standard IEEE 802.11p protocol of the VANET, which improves from 802.11a, the fading model is the multipath Rayleigh fading (Fig. 7).
Fig. 7

Vehicle movement model

In order to realize CGRP, we first implement the location service management protocol RLSMP. In the simulation, we simulate the scenarios of peak, intermittent and sparse traffic in a fixed area according to how many vehicles are distributed where the number of vehicles ranges from 160 to 600. Simulation parameters is in the Table 4.
Table 4

VanetMobiSim parameters and OPNET parameters

VanetMobiSim Description


Simulation range (km × km)

Number of lanes

Min speed (m/s)

Max speed (m/s)

Clusters density (vehicles/km)






OPNET Description


Beacon message size (bytes)

Beacon period (s)

Package size (bytes)

Transmission range (m)

Package generation rate (package/s)

Simulation time (s)







Figure 8 shows the simulation results of the end-to-end delay, routing hops and bit error rate of the four protocols with the increase of vehicles’ density. As can be seen from Fig. 8a, with the increase of the number of the vehicles, the delay tends to decrease, the probability that a vehicle carries a data packet will increase in low-density scenarios. Since the speed of vehicles is far lower than the speed of wireless transmission. In addition, in the case of high-density vehicles, the majority of messages use multi-channel wireless transmission. Since there are many vehicles with the same connectivity probability, the required communication range will be smaller, that is, the interference will be smaller, which greatly reduces end to end delay. Because GPCR and GPSR data packets are mostly transmitted in a carry-on mode, airborne wireless transmissions are much faster than vehicles travel, so they have a higher latency than CGRP. The OLSR can easily lead to routing failure, thereby having a higher latency.
Fig. 8

The relationship between the vehicles’ density and the performance of the network in the case of CGRP, GPCR, GPSR and the proposed OLSR. a The relationship between the vehicles’ density and the end to end delay. b The relationship between the vehicles’ density and the number of hops

Figure 8b shows the relationship between the forwarding number of the packet and the number of nodes in the four protocols. The result shows that GPSR has the least number of hops. Because it uses a greedy algorithm, it forwards the data packet to the nearest neighbor node to the destination node, so the number of hops is the least. The CGRP uses a priority algorithm which not only consider the distance, but also consider the speed, direction factors, so it has more hops than GPSR. The GPCR protocol, an improvement over the GPSR protocol, solves the problem of communication failures caused by vehicles at intersections. However, when a data packet is transmitted straight ahead, it may have been forwarded directly to the vehicle ahead of the intersection. However, according to the GPCR, When a straight packet is transmitted straight forward, the packet can be forwarded directly to the vehicle ahead of the intersection, however, according to the GPCR, it is still necessary to pass through the node at the crossroad, therefore, the number of unnecessarily jump and end to end delay increases. In addition, when the number of vehicles reaches a certain level, the selection of the next hop node will not change as the number of vehicles increases, that is, the number of hops will not change substantially.

5 Conclusion

In this paper, a connectivity-based multi-lane geographic routing protocol for vehicular communication networks is proposed. Firstly, the expression of connectivity probability of assistance in each lane is calculated based on the proposed three-lanes path model. Secondly, geographical forwarding is used to transfer packets between any two intersections on the path to reduce the path sensitivity. Finally, nodes will choose neighbor node with the highest priority to forward data packages by selecting the lanes with highest connectivity probability. Simulation results show that the proposed algorithm outperforms the exsiting routing protocols in terms of the end-to-end delay and the number of hops with a little cost of routing overhead in city environments.



This work was supported by National Natural Science Foundation of China (61771252, 61471203), Basic Research Program of Jiangsu Province (BK20171444), “The Six talents High Peaks” Funding Project of Jiangsu Province (DZXX-041), “1311”Talents Funding Project of Nanjing University of Posts and Telecommunications.


  1. 1.
    Blum, J., Eskandarian, A., & Hoffman, L. (2004). Challenges of inter vehicle ad hoc networks. IEEE Transactions on Intelligent Transportation System, 5(4), 347–351.CrossRefGoogle Scholar
  2. 2.
    He, R., Rutagemwa, H., & Shen, X. (2008). Differentiated Reliable Routing in Hybrid Vehicular Ad-Hoc Networks. In IEEE international conference on communications (ICC ‘08), (pp. 2353–2358).Google Scholar
  3. 3.
    Hrizi, F., Bonnet, C., Härri, J., & Filali, F. (2013). Adapting contention-based forwarding to urban vehicular topologies for active safety applications. Annals of telecommunications, 68(5–6), 267–285.CrossRefGoogle Scholar
  4. 4.
    Jiang, Y., Li, S., & Shamo, D. (2002). Development of vehicle platoon distribution models and simulation of platoon movements on indian rural corridors. Traffic Control, Res. Program 72.Google Scholar
  5. 5.
    Lochert, C., Hartenstein, H., Tian, J., & Fussler, H. (2003). A routing strategy for vehicular ad hoc networks in city environments. In IEEE intelligent vehicles symposium (pp. 156–161).Google Scholar
  6. 6.
    Naumov, V., & Gross, T. R. (2007). Connectivity-aware routing (car) in vehicular ad-hoc networks. In Proceedings of the IEEE INFOCOM (pp. 1919–1927).Google Scholar
  7. 7.
    Clausen, T., & Jacquet, P. (2003). RFC 3626: The optimized link-state routing protocol. Internet Engineering Task Force (IETF) Request For Comments.Google Scholar
  8. 8.
    Johnson, D. B., & Maltz, D. A. (1996). Dynamic source routing in ad hoc wireless networks. Mobile Computing, 353, 153–181.CrossRefGoogle Scholar
  9. 9.
    Lochert, C., Scheuermann, B., Wewetzer, C., Luebke, A., & Mauve, M. (2008). Data aggregation and roadside unit placement for a vanet traffic information system. In International workshop on vehicular ad hoc networks (Vanet 2008) (Vol. 11, pp. 49–65), San Francisco, California, USA.Google Scholar
  10. 10.
    R. Hussey, E. Huff, Z. Shinwari, and V. Hnatyshin, (2013). A comparative study of proactive and reactive geographical routing protocols for MANET. In 12th International conference on wireless network (ICWN’ 13), (pp. 1–6), Las Vegas, USA.Google Scholar
  11. 11.
    Li Zhiyuan. (2009). Geographic Routing Protocol and Simulation. In International workshop on computer science and engineering (Vol. 2, pp. 404–407).Google Scholar
  12. 12.
    Dua, A., Kumar, N., & Bawa, S. (2014). A systematic review on routing protocols for vehicular ad hoc networks. Vehicular Communications, 1(1), 33–52.CrossRefGoogle Scholar
  13. 13.
    Tsiachris, S., Koltsidas, G., & Pavlidou, F. N. (2013). Junction-based geographic routing algorithm for vehicular ad hoc networks. Wireless Personal Communications, 71(2), 955–973.CrossRefGoogle Scholar
  14. 14.
    Sermpezis, P., Koltsidas, G., & Pavlidou, F. N. (2013). Investigating a junction-based multipath source routing algorithm for vanets. IEEE Communications Letters, 17(17), 600–603.CrossRefGoogle Scholar
  15. 15.
    Chang, J. J., Li, Y. H., Liao, W., & Chang, I. C. (2012). Intersection-based routing for urban vehicular communications with traffic-light consideration. IEEE Wireless Communications, 19(1), 82–88.CrossRefGoogle Scholar
  16. 16.
    Karp, B., & Kung, H. T. (2005). GPSR: greedy perimeter stateless routing for wireless networks. In ACM International Conference on Mobile Computing and Networking (pp. 243–254).Google Scholar
  17. 17.
    Cinlar, E. (2013). Introduction to Stochastic Processes. Mineola: Courier Dover Publications.zbMATHGoogle Scholar
  18. 18.
    Saleet, H., Basir, O., Langar, R., & Boutaba, R. (2010). Region-based location-service-management protocol for vanets. IEEE Transactions on Vehicular Technology, 59(2), 917–931.CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Authors and Affiliations

  1. 1.College of Communication and Information EngineeringNanjing University of Posts and TelecommunicationsNanjingPeople’s Republic of China

Personalised recommendations