1 Introduction

Despite considerable improvements in transportation systems, using advanced equipment for navigation and dynamic allocation of the traffic, people still get stuck in traffic for a long time. With regards to the estimations, congestion costs until 2018 will be 192 billion dollars and fuel use will reach 3.8 billion gallon [1].

Hence finding a suitable solution for preventing congestion in urban environment and achieving lower travel time and fewer fuel consumption, is one of the essential challenges in this area. To prevent congestion, it is necessary to gather up-to-date traffic information from different urban environment zones and to offer this information to the drivers so that by using this information and making correct decisions, they can evade congestion in various regions of urban environment [2]. Systems, which prepare such information from the environment and offer it to the drivers, are called Traffic Information Systems (TIS) [3].

One type of these systems, in which information is collected from moving vehicles, are called cooperative TIS that are categorized into centralized and decentralized, according to the presence or absence of central management control center. In the former category (the centralized one), traffic information is gathered by means of vehicles and roadside equipment, then to be sent to the central management control unit. Results from processing such information are afterwards resent by communication network to moving vehicles [4, 5].

On the other hand, decentralized cooperative TIS are systems in which collection, distribution, analysis, storing, and searching traffic information is done by moving vehicles. For communication among the vehicles, these systems use inter-vehicle communications, infrastructure based communication or a mixture of both. Cooperative TIS usually employ a peer-to-peer layer on a physical one so that the actions of collection, saving, and retrieval of traffic information would be done with more efficiency. Such a layer can be built on the vehicles, infrastructure nodes, or both [3, 6].

Most presented cooperative TIS are centralized which, in accordance to the vastness of urban environment and high delay and costs of these systems, have low scalability. There is just a few which belong to the decentralized category that uses peer-to-peer networks on physical network for better efficiency. The peer-to-peer network in these systems is built either on vehicles or on infrastructure ones, assuming that a cellular network coverage exists all over the environment. In the former case (peer to peer on vehicles), due to the existence of weak inter-vehicle communication and continual network disconnection, such systems will not have the required efficiency, whereas in the latter case, establishment of an infrastructure that provides communication all over the urban environment will not be possible, owing to its unimaginable costs. Moreover, the assumption of a cellular infrastructure to establish information communication will not be efficient for traffic information exchange due to its high delays [7]. Hence it is necessary to have an efficient system in traffic information discovering and disseminating with suitable infrastructure costs and reasonable delay.

This article presents three Cooperative TIS by means of peer-to-peer networks. The first system only uses inter-vehicle communication; yet it is attempted to avoid network partitioning by suitably distributing the vehicle density, itself achieved by navigating the vehicles via different paths. Also in this system the packets are distributed in many paths with appropriate vehicle density as well, so that it will be less possible for them to be dropped by intermediate nodes as a result of limited network capacity, hence improving the efficiency. In the second system by deploying the RSUs in more suitable places of urban environment, the infrastructure-based peer to peer system will be used to collect and distribute traffic information. This system will not have the first one’s problems, as the connectivity of vehicular network will be guaranteed by RSUs; however, because of the high costs of installing and maintaining RSUs, covering all intersections by these units will turn out to be very expensive [8, 9]. Thus in the third system a subset of important zones is selected, in the most important of them a limited number of roadside units deployed, while in the other ones vehicles are used as peer-to-peer nodes. In case of installing infrastructure in the environment, the installation will not be uniform like other presented methods, but set up and deploy RSUs in the most suitable place in the environment in accordance to environmental and network limitations. What’s more in all three presented systems of this article, a peer-to-peer layer is established on physical layer. Thanks to this layer, collection, saving, and dissemination of traffic information is done more efficiently [10].

The key contributions of this article are as follows: (1) vehicle navigation and sending traffic information via separate paths by super peer nodes so that traffic information will not be dropped due to partitioning problem; (2) selection of many separate paths for vehicle navigation and sending traffic information so that the problem of congestion relocation and losing traffic packs will be resolved; (3) prioritizing entrance intersections of congestion zones and using infrastructure and vehicle as super peer nodes in intersections with high priority so that installing costs will be lessened and suitable delay will be achieved in terms of discovering and sending traffic information.

The rest of the article is organized as follows: It categorizes and studies the related work on cooperative TIS in part 2. Part 3 presents the proposed systems while part 4 evaluates these systems’ efficiency, comparing them with other existing systems. Finally, part 5 is dedicated to the article’s conclusion.

2 Related Work

Based on the use or no use of infrastructure nodes, presence or absence of peer-to-peer network, and presence or absence of centralized management control unit, cooperative TIS can be categorized as Centralized TIS [11,12,13] have high sending, receiving, and updating delays, thus they are not proper for sending and receiving traffic information in urban environment in large scales. Another type of these systems are the decentralized or distributed ones, in which inter-vehicle communications, infrastructure ones, or simultaneously both are used for discovering and sending traffic information. These systems are known as Single-Tier if only one network is used by them and Two-Tier in case two networks are mutually employed [7]. Also in these systems a second layer can be established on the physical network as a peer-to-peer network, in which case the created system would be two-layer system.

Of the introduced systems, the Single Tier VANET [14,15,16] is single-layered and the rest are two-layered, in which the second layer is the peer-to-peer network. Single-layered one-Tier system is the vehicular ad-hoc network or VANET. On this case the sent reports are aggregated and disseminated with the aim of saving network’s limited bandwidth. Traffic information is broadcasted with a low delay because of using direct inter-vehicle communication, yet efficient broadcast of the information requires proper density of the vehicles. Furthermore, in case of high vehicle density and broadcast of the messages by the vehicles, there arises the problem of broadcast storm.

In two-layered single-tier, a peer-to-peer network is formed on vehicular network in the application layer. The created peer-to-peer network can be either the structured p2p like Chord [17] or the unstructured one like Gnutella [18]. In these systems, the required traffic information is shared among the vehicles through the peer-to-peer layer, being received through the same layer as well. The main difference of this method with the normal wireless inter-vehicle network is in the way traffic information is searched. In this way, each vehicle does not broadcast the traffic information messages to all the other vehicles in its vicinity; rather by employing useable search methods in the peer-to-peer layer, traffic information messages are sent to the destination, instead of being broadcasted. Since in this system the peer-to-peer network is established on the mobile nodes, it faces the same problems of vehicular networks. In these systems the second layer is established either on mobile nodes [19, 20] or on the infrastructure ones [6, 21]; in the former case the system will result in high maintenance overload for peer-to-peer network, whereas in the latter case it suffers from long delay.

In another type of these systems both infrastructure and peer-to-peer layers are on infrastructure nodes and do not need any inter-vehicle communications. As a result, the partitioning problem of wireless inter-vehicle networks will make no problem here. The main problem of these systems is higher search delays, compared to the previous type. Presented system in Peer TIS [6] are of the same type. Another type of systems uses both wireless inter-vehicle networks and cellular infrastructure, in which the vehicles are initially organized in the lower layer and some of them are selected to form the second layer. The presented system in [3] belongs to this type. It should be mentioned that the created peer-to-peer layer in these systems are established either on mobile nodes or on the cellular infrastructure; in the first case it faces the problem of network partitioning and in the second case, long delay.

3 Proposed Peer to Peer Systems

Regarding Table 1, the main challenges in the field of cooperative TIS is as follows: (1) Absence of necessary connectivity in VANET single-tier systems that result in the inefficiency of such mechanisms in low penetration rate; (2) long delay in cellular-infrastructure-based systems that decreases these systems’ efficiency in timely presentation of traffic information to drivers; (3) sending packets with paths, the same as vehicles’ traffic paths which increases the challenge of network partitioning, because of vehicles’ navigation from sparse routes for lowering the travel time; and (4) vehicles’ navigation from a unique path that will periodically change the selected paths, letting the vehicles face roads congestion and car traffic, practically having no suitable influence on network efficiency and reduction of trip time.

Table 1 Comparing cooperative traffic information systems
Full size table

This article presents two-layer systems, which are based on peer-to-peer networks to improve TIS efficiency and resolve these problems.

For presenting these systems like Fig. 1, firstly the traffic files, taken from Zurich dataset, were evaluated and according to the given map, the environment got divided into some zones, among which the ones with congestion were determined by means of the datasets. In order to determine the zones with congestion, according to the given traffic files, for each zone the number of vehicles inside during certain time periods are measured, and eventually the average number of vehicles inside these districts are calculated. In case the number of cars within a area exceeds a certain limit, which is usually 70% of the zone’s vehicle capacity, the zone in question belongs to those with congestion.

Fig. 1
figure 1

Process of proposed systems creation and evaluation

Full size image

Usually, the directed traffic towards the zones with congestion is through certain paths; therefore, those intersections that are located in entrance paths into the congested zones are regarded as key points to redistribute and navigate the vehicles. Named congestion entrance points, these points can be prioritized, according to the number of road segments that have distance from congestion areas. In Fig. 2 points marked by +, are candidate points for placing super peer nodes. Among congestion entrances a subset of them are selected as chosen intersections which will be used in the systems, presented in this article, in order to create the peer-to-peer network.

Fig. 2
figure 2

Map from pre-process stage and determining congestion zones and entrance

Full size image

According to intersections’ priorities, a subcategory of those with high priority is determined in the pre-processing stage to select and establish super peer nodes in them. In each selected intersection, a node is chosen as the super peer node to keep the information about the status of the network’s congestion as well as the density of road segments and zones. Super peer nodes receive the information about travel time of road segments and the number of vehicles in each road segment from the neighbor vehicles, and update the related information. In fact, each super peer keeps a graph of the network in itself, in which the intersections are regarded as vertices and the road segments between them as the edges. Each created graph edge has two different weights, one to show the average travel time in that road segment and the other the connectivity probability of the road segment, itself. Eventually the most suitable path for vehicle’ navigation is selected from the created graphs on super peer nodes, and the suitable path is calculated for traffic packet dissemination.

In these systems four kinds of packets are generated by the nodes: (1) normal packet, to exchange information among moving vehicles in same road segment; (2) super peer packet, to exchange information among second layer peers in differing road segments; (3) search packet, to search and update the information of travel time for road segments, suggested by systems for vehicle movement; and (4) response packet, which are produced in response to search packets by vehicles, having the required information in search packet.

For gaining the capability of updating the network status, each node keeps three traffic information tables, the fields of ea are like Table 2 as follows: (1) Table for saving travel time of road segments, (2) list of neighbor moving vehicles of the same road segment in underlay layer, and (3) table for the information about the probability of connectivity of road segment in underlay layer.

Table 2 Tables of traffic information, saved in each vehicle
Full size table

The table for the data of road segments’ travel time, is created based on information, prepared in a cooperative manner by the vehicles, and is used to determine the best path for vehicles’ movement. Table for the data about the probability of road segments’ connectivity is employed to send search and response packets in peer-to-peer layer so that these packets will be sent to their destinations from the most trustable paths in underlay layer with suitable connectivity probability. Furthermore, table for first layer neighbors has two uses: (1) to estimate the probability of road segments’ connectivity, and (2) to determine the second layer’s peers. All vehicles in a road segment are defined as a cluster. Each time the road segment is changed by the vehicle, adjacency table’s data are deleted and a new road segment are updated, based on the neighbors’ information; yet the data of the other two tables remain and will be updated only by receiving new information.

Each vehicle generates all three tables in each of the normal packets, which are themselves created periodically by all nodes in periods of time that alter in accordance to the vehicle’s speed. Vehicles update their tables based on the information they experience in their movement as well as with the one they receive from other vehicles about various road segments. Each node aggregates the new information of each road segment with its previous information, saving it with a time stamp. In order to update the information of both tables of time of passing the road segments and information of road segments’ connectivity, one can use Formula 1:

$$ T_{Agg} = \alpha T_{Old} + \beta T_{New} $$
(1)

where TAgg is aggregated travel time of related road segment that will be stored instead of old data, TOld is currently stored travel time of the road segment; TNew is the new received travel time of that road segment. Coefficients α and β are defined as formula 2:

$$ \alpha = 1/\left( {1 \, + \Delta T} \right); \, \;\;\beta = \Delta T/\left( {1 + \Delta T} \right) $$
(2)

In Formula 2, where ΔT = TNew − TOld and α + β = 1. This formula controls the updating rate of traffic information in Travel-time table. It means that when the difference between times of last updated and time of receiving new data is too long, the new one has much more weight than the old one. Therefore, the table information will become freshen with the new data. On the other hand, when a node receives some similar data about a road segment in a short period of time, this formula try to control the effect of this new data by assigning a low weight to them by a small ΔT. These two advantages may prevent navigating part of the algorithm to plan according to very old or much fluctuating information. To estimate the probability of road segment’s connectivity, each vehicle uses its saved adjacency table while moving. Based on its position and that of its neighbors, which are included in the adjacency table in either direction, it calculates the probability of connectivity of the road segment, in which it is located. This calculation can be made as follows: if the n vehicles exist in both directions in a node’s adjacency table, by means of these vehicles as well as the same node, the road segment can be divided into n + 2 sub-segments, created by consecutive vehicles. The probability of this road segment’s connectivity can be then calculated via formula 3:

$$ P_{Connectivity} = {{\left( {\sum\limits_{i = 1}^{n + 2} {d_{i} } } \right)} \mathord{\left/ {\vphantom {{\left( {\sum\limits_{i = 1}^{n + 2} {d_{i} } } \right)} L}} \right. \kern-0pt} L} $$
(3)

In which PConnectivity is probability of the road segment’s connectivity; L, its length (in meters); and the di are connectivity length of road segment (in meters) and could be calculated through formula 4:

$$ d_{i} = \left\{ {\begin{array}{*{20}l} d \hfill & {d < R} \hfill \\ R \hfill & {d \ge R} \hfill \\ \end{array} } \right. $$
(4)

where R (in meters) is the maximum range of vehicles’ sending and receiving, adjusted in the system by default (for example 250 m) and d (in meters) the distance between two consecutive vehicles or between vehicles at the end of the road segment and the intersection, itself.

Also the probability of a complete path’s connectivity, from the beginning to the end, can be calculated by means of the data table of connectivity probability, saved in each vehicle, as well as using formula 5:

$$ P_{Path} = {{\left( {\sum\limits_{i = 1}^{m} {P_{i} } L_{i} } \right)} \mathord{\left/ {\vphantom {{\left( {\sum\limits_{i = 1}^{m} {P_{i} } L_{i} } \right)} {\left( {\sum\limits_{i = 1}^{m} \begin{aligned} \hfill \\ \hfill \\ \end{aligned} L_{i} } \right)}}} \right. \kern-0pt} {\left( {\sum\limits_{i = 1}^{m} \begin{aligned} \hfill \\ \hfill \\ \end{aligned} L_{i} } \right)}} $$
(5)

In which PPath is the probability of a complete path’s connectivity from the beginning to the end, being comprised of m connected road segment. The Pi shows th probability of each saved road segment’s connectivity, based on its related table and the Li (in meters) are the corresponding length of that road segment. Formulas 3 and 4 are employed to update the table of probability of road segments’ connectivity and formula 5 is used for search packets’ navigation through the paths with the most suitable rates of connectivity probability.

As a key problem of VANETs is the limited bandwidth for sending packets. Therefore, it is useful to control the rate of sending packets in such a way that prevent reaching the saturation thresholds. In our proposed system, there are two periodic broadcasting packets: Normal and Super-peer packets. To save the VANET bandwidth from sending useless data, the algorithm tries to transmit these two types of packets in adaptive intervals depending on the node’s speed. It means that instead of transmitting these two packets in fixed intervals, each fast moving node can broadcast Normal and Super-peer packets more frequent than slow moving nodes using (6):

$$ {\text{Ia}} = {\text{I}}0\left| {40{-}{\text{v}}/{\text{V}}0} \right| $$
(6)

where Ia is broadcasting interval for sending two consecutive Normal or Super-peer packets (sec), I0 is a predefined interval for beacon sending in VANETs and is considered 0.1 s, v is node’s speed (m/s) and V0 is a reference speed (m/s) to control the change rate of the formula and is considered as 1 in this round of our project. Constant value of 40 has role of controlling minimum time interval in (3) according to this fact that the maximum valid speed of city roads is less than 40 m/s (144 km/h). By using (6), fast moving cars can broadcast in shorter intervals than stopped or slow moving nodes.

Below, each presented system will be surveyed to show the procedure for both the selection mechanism of super peer nodes and placement of roadside units in infrastructure-based systems.

3.1 Vehicles’ Navigation and Traffic Packets’ Dissemination via Peer-to-Peer Networks on Mobile Nodes

In this case it is assumed that there is no infrastructure node in urban environment, thus like the presented solutions, not using the infrastructure, it suffers from network partitioning in low penetration rate. All the same, based on the diagram given in Fig. 1 as well as the cases below, the system will show suitable efficiency, compared to other solutions. Among the vehicles, near the chosen intersection, the system selects the nearest, hence reducing the overload of super peer nodes selecting.

By means of the packets, introduced in Table 2, the information about road segments and their travel time is given to the nodes that reach the intersection. By using the algorithm of the shortest path, each node determines a total of K more appropriate paths for arriving at the destination. Among the K more suitable paths in terms of travel time, the vehicle chooses one in random, moving through it towards the destination (the next chosen intersection). Such an action distributes the vehicles, which have arrived at the intersections, among the K better paths, so that the sparse zones will not face vehicle congestion rapidly.

Apart from the vehicles, sending traffic packets is done with a similar mechanism of vehicle navigation. In this case too, the created graph is used to determine the most appropriate path for sending traffic packets and the paths, with low probability of partitioning, have to be selected. Moreover, such paths should not be located in zones with congestion, since in such zones due to shortage of network resources the probability of losing the packet increase; as a result, in this case too, the packets are distributed among the K paths with suitable vehicle density.

Although this method does not use roadside infrastructure, it receives help from a peer-to-peer network with low overload in the second layer and suitably distributes the vehicles and traffic packets, hence improving the efficiency of vehicular networks, trustable dissemination of the packets, and decrease of vehicles’ travel time in urban environment.

3.2 Vehicles’ Navigation and Traffic Packets’ Dissemination via Peer-to-Peer Networks on Infrastructure Nodes

Since one of the main reason of the previous system’s limited efficiency was network partitioning in low penetration rate and high costs to update the peer-to-peer network, in the one, presented here this problem is solved by appropriately installing the infrastructure and establishing a peer-to-peer network on it. Infrastructure is deployed based on the priorities of congestion entrance intersections as well as the selection of the most important intersection for installing roadside units.

To determine the intersections in which these units should be installed, every intersection is given a priority. When prioritizing the intersections, it should be taken into consideration that those in the entrances of congestion zones and have low vehicle density are in the first priority for installing. Moreover, those that need no coverage or are not suitable for installing the RSUs are omitted from the prioritization list.

The total costs of installing in each zone or intersection can be calculated based on Formula 7:

$$ G_{i,j} = C_{i,j} + \beta L_{ij} + \alpha N_{ij} D_{ij} $$
(7)

According to Formula 7, if an intersection has average vehicle density, making inter-vehicle communications possible, the installing costs will be the costs of installing the RSU in the intersection in addition to the extra costs with a coefficient of α, which makes total installing cost become a huge number; therefore, this intersection will be removed from priorities due to its high costs. Furthermore, if an intersection has no vehicle traffic, needing no coverage by RSU, the number will be again a huge one. If the intersection’s distance from the congestion area is small, the resulted cost for installing the RSU will be small, increasing the chance of installing a RSU in the intersection. Based on Formula 7, every existing intersection is given a cost in which this cost will be smaller for those near the congestion area, thus with increased probability of installing roadside units in these zones.

At first, the costs of installing the unit is calculated for each zone, based on Formula 7. By means of the calculated costs, as long as all zones, in need of coverage, are not covered by RSUs, they are placed in coverage-needing zone which has minimum cost of installing. Also in every stage it is studied whether the intended zones are covered by the RSUs, already installed in other zones, or not. This avoids multiple coverage of the zones, minimizing the number of installed RSUs in the environment. In fact it is intended to keep the minimum number of roadside units within the environment. Hence the costs of installing the units in the environment will be minimum. Basic idea of this algorithm is based on [8] in which the deployment strategy is optimized for minimizing the number of RSU and overlap zones with lower time complexity. Moreover, some modification is done to improve it’s accuracy in large scale. Used symbols in presented algorithm, are shown in Table 3.

Table 3 Used symbols in deployment algorithms
Full size table

In this case, by installing roadside units as the infrastructure node, the problem of network partitioning is solved. Moreover, since these units are regarded as super peer nodes, the updating costs of the network will be reduced as well. As super peer nodes, these units update the network graphs by receiving traffic information from the vehicles that arrive at the zones; and by providing these updated data to the arrived vehicles, help them choose the most appropriate travelling path and the best one to send traffic information. The main challenge of this system is the high cost of deployment roadside units in the environment, which is solved in the next system by presenting a hybrid solution.

figure c

3.3 Improving Vehicle Navigation and Dissemination of Traffic Packets via Peer to Peer Network on Both Infrastructure and Mobile Nodes

Assuming the presence of an infrastructure solves the problems, arising from vehicles’ low penetration rate in urban environment; however, considering the high costs of installing and maintaining the infrastructure, there should be a limited number of roadside units, installed in the environment. As a result, finding the optimal locations for deploying limited number of RSUs will be of high importance. Unlike the second system, in this case there are just a limited number of roadside units, which in accordance to priorities of the zones that are located in congestion entrances will be installed with higher priority. If congestion entrance intersections are not covered by roadside units, like the first system, the nearest vehicle to the intersection in each will be selected as super peer node, hence in this system the peer-to-peer network is established on the infrastructure node, installed in the more important intersections of the selected zones, along with in the selected node in the intersections of congestion entrance as super peer mode.

It can be said that this system is a tradeoff between the costs of installing and maintaining RSUs and the required environment coverage by means of these roadside units, themselves. In this system, like the previous ones, there are K separate paths for navigating the vehicles and disseminating the packets by the nodes.

This system often solves the mentioned problems and challenges as it meets the problem of environment’s partitioning by having a limited number of roadside units while it charges limited costs for installing the infrastructure. Furthermore, by separately navigating the vehicles and disseminating the packets, it reduces the probability of losing traffic information packets, lessening the vehicles’ travel time as well. Additionally, the created peer-to-peer system has lower overload of generating and updating, compared to the first system.

4 Performance Evaluation and Simulation Results of the Proposed Systems

This section will deal with evaluating the efficiency of the proposed systems. Table 4 gives simulation parameters along with the tools to simulate the proposed systems. Different parameters have been measured when evaluating the systems that include vehicles’ average travel time in different systems, in accordance to various penetration rates as well as the influence of vehicle density on average travel time.

Table 4 -Simulation assumptions
Full size table

Based on Fig. 3, it can be seen that the more the number of vehicles, the long the travel time; nevertheless, the presented systems decrease travel time in different densities. Two more influential parameters of the presented systems’ efficiency are the number of separate paths, selected by the systems, and the distribution of the vehicles and traffic packets among these paths.

Fig. 3
figure 3

Comparing of different structures based on avg. travel time in various densities

Full size image

In case of only one path, being selected for so doing, it will face congestion intermittently, thus the vehicles travel through other paths in time. Periodical changes of the paths reduce the proposed system’s efficiency, because prior to vehicle re-routing, some will get stuck in the zones with congestion. In these systems, many paths will be selected instead of one, vehicles and traffic packets will be directed through these paths, and the chances of paths facing congestion will be lessened. Although increasing the number of alternative paths will decrease the travel time, processing overload of peer-to-peer nodes will increase, considering the calculation of separate paths. Figure 4 illustrate the influence of the number of selected paths in each peer-to-peer node on the average travel time. It can be seen that increasing the number of paths for vehicle and traffic packets distribution, decreases the average travel time as well as vehicle congestion.

Fig. 4
figure 4

Avg. travel times versus number of different path

Full size image

Figure 5 depicts the influence of traffic information storage by the vehicles and infrastructure nodes on lowering travel time. As it can be seen, by increasing the penetration rate, the influence of storing traffic information will become more significant on lowering travel time. Yet if the rate of storing traffic information exceeds a certain limit, the efficiency of these systems will be reduced, considering no update of these traffic data and the probability of their invalidity. Yet, since saving the messages, reduces the number of sent/received messages, the presented systems act more efficiently in terms of the network’s used bandwidth by saving them.

Fig. 5
figure 5

Average travel time with regards to penetration rate and various storage time (in seconds)

Full size image

Figure 6 shows how storing the messages, decrease the number of sent/received messages, and eventually the network’s required bandwidth. As it is shown in Fig. 6, by increasing the storage time of the messages, the number of transmitted ones decrease; all the same, a storage time of beyond 600 s reduces the system’s accuracy since saved outdated information in the nodes do not have the necessary accuracy.

Fig. 6
figure 6

Number of sent/received messages with regards to their different storage time

Full size image

5 Conclusion

Based on the simulations and their results, it can be said that the systems, presented in this article, reduce vehicle congestion and average travel time. Even in low penetration rate, the presented systems show suitable efficiency. By increasing the vehicles’ penetration rate, both travel time and congestion will decrease. Moreover, message storage in these systems saves the vehicle networks’ limited bandwidth; additionally, using many paths in vehicle navigation and traffic packets forwarding avoids congestion and travel time decrease via suitable distribution of the packets and vehicles. Furthermore, with this method, the sent traffic packets will not be forwarded to paths with high congestion, in which they may be dropped due to the shortage of network bandwidth. What is more, deploying the infrastructure at appropriate points reduces vehicles’ travel time and improves network’s connectivity. Considering high costs of the infrastructure, the article has given a hybrid system, the peer-to-peer nodes of system consist of infrastructure and mobile nodes. Having suitable efficiency, it solves most of the mentioned problems of the previous systems.