This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Vehicular Ad hoc networks is a special type of Mobile Ad-hoc Networks in which the moving vehicles act as nodes. In VANETS communication can be carried out between vehicle to vehicle (V2V) or between vehicles to infrastructure (V2I). In vehicle to infrastructure, vehicles communicate with road side units as shown in the figure 2.1. Further, MANETs have many challenging tasks like unpredictable topology and can change more frequently from time to time. Also the Network is not centralized and no access point is involved in MANET to control the whole network. Vehicular Ad-hoc networks (VANETS) is more challenging as compared to the MANETs in these sorts of environments because of high mobility of vehicles as compared to the nodes in MANETs. The network topology in VANET changes more frequently as compared to the MANET hence VANETs are highly dynamic in nature.
Figure 2.1 VANET Communications
Zang et al.  computed Vehicular network neighbor vehicle communication in which he used 4 lanes with each direction having 2 lanes. Through simulation it was computed that the neighbor vehicles which were the part of effective communication having an average of 4. Further, in almost 50% events, communication duration of effective communication is one second only. In 90% of the events, the coordination duration of the vehicles was 5 second for the upper boundary.
There is another important factor in vehicular ad-hoc networks (VANETS) is the lack of bandwidth. In these sorts of environments the transmission channels are occupied by different nodes e.g due to collisions occurred among the nodes, interferences between the nodes and the wireless channels , signal strength can be weak, time required for communication sequence, etc. .
To cope with these sorts of challenging task in these environments and to provide reliable communication and safety measures, setup of best policies based on quality of service is compulsory, which inspires congestion control in VANETS Figure 2.2 (c.f Figure2 .)
Figure 2.2 Congestion Control in Vanet
To control the congestion in VANETs is a challenging task due to the high mobility of nodes and dynamic behavior of topology. Because the transmission channel is by default shared in nature, the chances of the congestion are high. Links are broken more frequently in VANETs due to high mobility as well as the routes which can increase latency in the network and can increase packet loss ratio. Moreover only single communication is possible in shared transmission channel, hence congestion can effect badly on overall Mobile ad-hoc networks .
Many Algorithms have been proposed so far to control the congestion in VANETs. In end-to-end Congestion control mechanism Chandran et al.  proposed by using an idea of TCP Feedback or called as TCP-F, In this technique when the error occurs in routing or the the routes are changed, this information is passed immediately to the source node so that it can freeze the state values of TCP like contention window size and timers etc.
End-to-End congestion control is also proposed by Rath et al.  using TCP and layer 1. Because the window based congestion scheme used by wired networks may not be suitable for wireless networks because of the communication nature is different in both wired and wireless networks, also in wireless networks interference can cause latency in the network and the packet loss. TCP Congestion control scheme has no idea whether the congestion is due to packet loss or the interference, it applies the same scheme in both cases. Hence due to this MAC link layer alters the communication power according to the nature of wireless channel and the interference of the vehicles, on the other hand TCP layer handles congestion by using Reno-2 approach.
Two new congestion control mechanisms are proposed in  using the flow control TCP model. One algorithm is static and the second one is dynamic algorithm. Working of both these algorithms is to first check how many connections a node have in the network and then alter the no. of connections in an appropriate way as per network requirement.
In Hop-by-hop Congestion control mechanism has been discussed in , that the congestion control mechanism schemes that are being used in wired networks cannot be appropriate for wireless networks because of dynamic nature. Hence these techniques should be re-designed. Because of the broadcast nature of wireless networks congestion acknowledgment can be get from each link which are directly connected. But the adjacent links which are not connected in a direct manner have the chances to interfere.
Some arguments were made in  by Yi et al. that hop-by-hop mechanism of congestion control can be utilized in wireless communication. In this technique when the congestion occurs the preceding node is notified about it, then the preceding node adapts its communication rate with this acknowledgment. This acknowledgment is usually provided with the queue length of the congested node. When the threshold is reached at the length of the queue then the preceding node is informed about the congestion so that it can decrease its transmission rate to avoid any further congestion. By using this technique congestion is limited and it can provide bandwidth effectively. This proposed algorithm works on the queue length and congestion acknowledgement, but it does not provide the forwarded data validity, therefore this method was not considered as the best method to adopt.
To control the congestion C3TP  was proposed which computes the channel capacity and allocates the data rate with these measurements. This technique uses an extra module which will compute these tasks. Similar method has been used by Rangwala et al.  in which distributed rate controller was used to compute wireless channel capacity and allocate data transfer rate to reduce the congestion.
Zang et al. in  proposed Congestion Control in wireless network for vehicular safety applications. The main purpose of this technique is to handle congestion by event-driven detection and to make sure the reliable transfer of safety messages by using Quality of service. This method is invoked at the run time to distinguish priority message for reliable delivery. Computation is carried out by considering wireless channel capacity. Effective packet forwarding optimization is not achieved because channel bandwidth sharing parameters are not evaluated. L. Wischhof and H. Rohling  presented congestion control mechanism for vehicular adhoc networks using hop-by-hop method. This method also provides efficient packet forwarding optimization in VANETs. The packet which needs to be transmitted is encrypted in a transparent way, Later on, an algorithm is applied which calculates the Average utility value of each vehicle. This method uses periodic exchange of messages among vehicles, which increases network overhead . Another issue in this method is that by computing messaging priority by means of data packet size can decrease disseminating performance of event-driven safety messages.
The congestion control mechanism in VANETs  are proposed by using the QoS parameters with bandwidth measurements. Hence vehicles exchanging priority information will be carried out first to utilize the bandwidth as per requirement. The priority is measured for each data packet then based on its measurement the transmission data rate is calculated. This method can cause transmission overhead because of the information being exchanged among vehicles for sharing the bandwidth.
Torrent-Moreno et al.  proposed an ideal sharing of bandwidth in Vehicular Adhoc networks. This method reduces the load which are carried out by extra information being exchange among vehicles. A centralized power control algorithm is proposed to provide communication range for each vehicles as per their requirement. This method needed periodic synchronization among nodes that can create network overhead.
Another proactive Congestion control mechanism is presented by M.S BOUASSIDA and M. SHAWKY  which is called Dynamic Priority- Based Scheduling (DPBS). This scheme overcomes the disadvantages in hop-by-hop methods. This method provides reliable and efficient packet forwarding techniques and make sure to transfer the packet from source to destination in an efficient manner and based on dynamic priority based scheduling. In this scheme the data with highest priority is sent first. Also and medium and low priority data are rescheduled. There are two types of scheduling process one is static and the other is dynamic. Static process evaluates every message priority and align in the transmission channel queues. Dynamic process on the other hand periodically evaluates messages in the queue and computes the total priority. When the Transmission channel is available for communication, highest priority data is sent first, forwarding of low priority messages will be paused so that high priority data is transmitted first without any delay.
Kharokhun, et al.  proposed an algorithm based on reactive congestion control, the idea was to configure transmission power or packet transmission rate by calculating ratio of busy time wireless channel. Where the busy time is the total time duration in which the wireless channel remain busy. Packet transmission rate or transmission power can be increased or decreased. For best performance it is mandatory to pass these measurements between nodes and increase the transmission power or packet transmission rate only if current value being used is less than the average transmission rate or power of neighbor nodes. As compared with proactive congestion control techniques, the reactive congestion control technique is not suitable for wireless channels to prevent congestion and it does not considers quality of service for message priorities for data transmission. The problem arises when each node evaluates transmission power in the same time, neighbor vehicles can monitor channel busy time has been reduced due to decrease in transmit power and those neighbor vehicles have not decreased their own transmission power. Hence, these neighbor nodes will increase their transmission power which should be decreased in this scenario. Hence some improvements should be made to avoid this problem i-e which one should decrease transmit power first and which one should decrease later on. Moreover it should be identified the reason of decreased wireless channel busy time.
A Hybrid Congestion Control Mechanism takes the advantages of proactive and reactive congestion control mechanism, presented by Baldessari et al  based on rate congestion control and combined transmit power. This algorithm evaluates the channel busy time and find out the no. of neighbor nodes around area and can provide efficient rate control, better transmit power control and also combined transmit power and transmit rate control algorithm. Depending upon no. of neighbor nodes a threshold is defined for channel busy time, packet generation rate is computed which will not increase from the predefined threshold value. Later on nodes are distributed uniformly using proactive Congestion method.
Due to highly dynamic nature, Routing in VANETs is a difficult task. Hence routing in VANETs becomes complex. Ad hoc on Demand Distance Vector routing protocol (AODV)  is an on-demand routing protocol used in VANETs. AODV is a type of reactive routing protocol. Every nodes in the wireless networks behaves as host and router. Routes are maintained in the form of routing tables and every node should have routing table to forward the packet towards the destination. The data packet is transmitted using the states of route request message (RREQ), route reply message (RREP), route error message (RERR) and Hello packets for connection validity confirmation.
When a node wants to forward the packet towards the destination it broadcasts the message in the form of RREQ. If the receiving node is not the destination it re broadcasts RREQ further until the packet reaches its destination. Upon receiving the destination node send back RREP to the source node. When routing error occurs or link is broken RERR is generated.
Congestion can occur in AODV routing protocol due to excessive broadcast. AODV can avoid congestion only when the sending node does not receive RREP from its intended recipient it re broadcast RREQ after some maximum RREQ_RETRIES. There is another way in which AODV can handle congestion is to decrease the transmission rate when congestion is likely to occur in the network and scheduling the network so that it does not become chock. The Disadvantages of AODV  is that it uses routing table and since the VANET topology changes more frequently it can make the network overhead. Secondly the connectivity between nodes breaks more frequently due to high mobility of nodes. Also AODV routing protocol can create congestion due to exchange of routing tables between neighbors.
Initially AODV was designed to be used for MANETs scenario, but AODV was not suitable for VANETs because of high mobility of vehicles. Later on DAODV  routing protocol was designed to be used with VANETs. The working of DAODV in is same as AODV, however DAODV also consider Direction of the vehicle to transmit the packet in the appropriate direction which is a very important factor in VANETs scenario. When a node wants to forward the packet to its destination the source node send RREQ route request message to the selected nodes depending upon the direction and movement and this message is sent in the form of multicasting. DAODV uses routing tables to forward the packet. Routing tables are exchanged periodically between neighbors. Since the network topology changes rapidly in VANETs routing table updating and maintaining are done frequently.
When the receiving node receives RREQ message, if the receiver is not the destination it then re-multicast that message to other nodes. When the message is received by the Destination it simply send RREP message back to the source.
If the packet during the whole transmission is dropped or destroyed due to any reason a route error message (RERR) is generated and sent to the source node. After receiving RERR message source node again starts the process. The Main Disadvantages of DAODV is that it can make the network overhead due to frequently change in topology. The ratio of connectivity break is high and it can create congestion due to exchange of routing tables.
Geographic Source Routing in  was initially designed for MANETs. After that it started being used for VANETs by improving it further. This protocol was improved by including greedy forwarding of information from source to destination. If no nodes exist between source and destination GSR uses recovery mechanism. This mechanism is called as perimeter mode. The perimeter mode consists of two algorithms. One is called distributed polarization algorithm. This algorithm helps to remove redundant edges by using local converting of connectivity graph to planar graph. The other Algorithm is called online routing algorithm. This algorithm works with planer graphs.
Geographic source routing uses perimeter mode in VANETs. It forwards the packet to the adjacent nodes instead of forwarding it to the farthest node. Due to large number of hop counts latency in this algorithm can be high. Routing loops can occur due to high mobility of vehicles that can create data dissemination to long path. Static street map is used in GSR and it has location information of every node in the network. GSR is not considering being an efficient method because it does not consider vehicle density. The Drawbacks in GSR is High latency because many number of hops involved. In low traffic density end-to-end connection establishment is difficult, Moreover in highly dynamic topology end to end connection scheme cannot work.
Greedy perimeter coordinator routing  works with city scenarios issues. It uses restricted greedy forwarding procedure in a predefined path. After selecting next hop, a coordinator is given preference over non coordinator node. The coordinator should not be the nearest node to the destination node geographically. The Drawbacks are Requirement of Static street map is compulsory and in low traffic density end to end connection scheme is a challenging task
A-STAR (Anchor-based Street and Traffic Aware Routing protocol  was designed to establish a reliable end to end connection in VANET which has low traffic density. In this protocol the numbers of junctions are calculated towards the destination. This protocol finds the path by using street awareness and traffic information. Hence static and dynamic maps help to computer number of junctions. In Static maps, high connectivity is achieved by using schedule of buses. E.g. in some areas regular street buses provide services and the buses connectivity can be high. On the other hand in dynamic maps, in order to compute junctions to find the path fresh traffic information is collected. Some roads are broader and thus have high traffic movement. Hence broader roads have high traffic as compared to less wider roads. A-Star protocol assigns weight to street by using this sort of traffic information. Itâ€™s a dynamic process which helps to computer anchors much effectively. It has certain disadvantages like it has lower packet delivery ratio as compared to GSR and GPSR and it is dependent upon static maps to calculate path from source node to destination that can create connectivity issue in some areas.
MDDV Mobile-Centric Data Dissemination Algorithm for Vehicular Networks  was designed to provide better routing scheme more effectively despite high mobility. MDDV combines the idea of opportunistic forwarding, trajectory based forwarding and geographical forwarding. In MDDV packets are sent to the destination in a predefined trajectory geographically. Intermediate nodes must buffer and send the data opportunistically. The selection of forwarding trajectory depends upon traffic density. The nodes have the knowledge of topology by using digital map and its own location by using GPS device. Its Drawback is that
Trajectory based forwarding does not work with high mobility of nodes which is the main disadvantage of this protocol.
VADD Vehicle-Assisted Data Delivery  routing protocol is used in those networks which have less delay and it make sure to establish a reliable connection. VADD Protocol uses carry and forward technique. If there are multiple paths towards the destination then VADD protocol selects those paths which have higher speed. It is a routing loop free algorithm and if routing loops are detected it immediately recovers from it. It forwards the packet by selecting nearest node towards the destination. VADD has many disadvantages, since the topology changes rapidly large delay can occurs. Large amount of bandwidth is consumed by exchanging hello packets. If the routing table has old routes it can cause network to become unstable. Since vehicles have high speed it does not operate well.
DGRP Directional Greedy Routing protocol  is based on greedy routing protocol. The selection of the next hop is based on the criteria which use geographic location, vehicular movement and their positions. It continuously evaluates the location and direction of the vehicles and it delivers the packet when possible. This is because DGRP already have the information regarding vehicular movement, mobility and position. Moreover, average loss of the data is much higher in DGRP if the connection between source and destination is unstable. The Drawbacks are average loss of data ratio is high, Latency in dense network and uses large number of hops to forward the packet.
Greedy Perimeter Stateless routing protocol GPSR  is to provide scalability by using geographic location. The purpose of considering scalability in this routing protocol is that it can give better performance in high traffic movement areas and high mobility of vehicles. There are two Algorithm used in GPSR one is called greedy forwarding that can be used at any time while other is called perimeter forwarding which works where greedy forwarding is not applicable. The main advantage of GPSR is that it knows only the information of forwarding vehicles. This protocol depends upon the vehicle density of the network. It does not rely upon how much destination nodes are present in the VANET. The Disadvantages includes Latency that can be high because it uses large distance to forward the packet. It is only appropriate for dense network, otherwise it may not work properly.
Jiayu Gong in  proposed PDGR (Predictive Directional Greedy Routing). Two strategies are used which are called position first forwarding and direction first forwarding in which weighted score is computed. The criteria to select next hop is based on the prediction but is not suitable in every kind of situation. It does not make sure the data is forwarding to the vehicle available at the Transmission range edge, which is assumed to be the most reliable next hop due to high dynamic nature of the nodes. Hence packet loss ratio and latency can occur in this protocol. Also it can increase routing overhead.
The protocol in  Potential Edge Node Based Greedy Routing Algorithm (EBGR) is designed that any vehicle in the network can send the packet to any other vehicle which is also called unicast message forwarding. In other case one vehicle can transmit packet to all other vehicles in a network which is called as broadcasting. The main purpose of this protocol is to provide reliability and it guarantees to deliver the packet in high mobility environment. EBGR has following three strategies.
Neighbor Node Identification strategy.
Node Direction Identification strategy
Edge node Selection strategy
In neighbor node identification strategy whole information is gathered of all the nodes available between source and destination. In Node Direction Identification strategy is to gather information of all the nodes which are heading towards the destination. In Edge node selection strategy transmitting of the packet is based on the selecting a specific node on the edge within the communication range. This strategy does not work when the nodes are moving with different speed because The edge node selection unit of this protocol evaluates the edge nodes that does not lies in between the transmission range.
Fisheye state routing  is a type of proactive routing protocol which computes the routing table by getting all the nodes information in the network from the neighbors. Fisheye state routing works on the idea of link state routing protocol. Since FSR exchanges information with neighbors only hence it uses less amount of bandwidth, also only the partial information is exchanged between neighbors not the whole routing information thus it has significant advantage over bandwidth utilization. Routing tables are only updated if there is a change occurs in the network hence routing overhead are reduces. The Drawbacks are it does not perform well in large networks. In large networks processing increases so as the routing overhead. It knows information of neighbors only and it does not know about farthest nodes and does not have complete information in the routing table.
Temporally Ordered routing protocol  forms acyclic direct graph in the direction of destination in which tree root is the source node and it works on the link reversal algorithm. In this protocol sender broadcasts and when the packet is received by the other node then it rebroadcasts the packet. TORA forms direct acyclic graph whenever needed. All the nodes in the network do not rebroadcast the information thus eliminating overhead in the network. TORA also has much better performance where network density is high. But routing table maintenance is a challenging task in VANET. Also some other routing protocols have much better performance as compared to TORA. Scalability is not achieved in TORA.
Greedy Traffic Aware Routing (GyTAR)  is based on intersection routing protocol in which network have fewer control message and it also helps to overcome losses in packets and latency in network as well. This protocol depends upon RSU units which have many disadvantages
Connectivity Aware routing protocol CAR  is best suitable for vehicle to vehicle communication. It helps to find the destination which based on links from one node to another. CAR uses certain predefined criteria to find best path from source to destination. PDR ratio is high in CAR. But it has some limitations like useless and extra nodes can be elected as anchors.
Street Topology Routing protocol (STBR)  has three components. One is called master, other is called slave and the third one is called forwarder. One master node is chosen while the rest of the nodes behaves as slaves while the in between nodes of junctions behaves as forwarders. When one node gets out of the junction then shifting of the routing table to newly elected master node is a challenging task.
GRANT Greedy Routing with Abstract Neighbor Table routing protocol  is based on the idea of greedy routing scheme which helps to prevent local maximum Greedy Routing. The planar areas are divided and only neighbor per area is selected. On the streets which have dense traffic and high buildings etc. which can prevent communications etc. this routing protocol is best suitable for that. But frequent exchange of hello packets can cause the network to become overhead and performance of GRANT is measured on static traces.
In Cluster Based Location Routing (CBLR)  two or more nodes in a network forms a cluster. There is one cluster-head in each cluster managing all activities in the network. In intra cluster connectivity of every cluster is by using direct connectivity whereas in in the case of inter-cluster connectivity is done through cluster head. Only the cluster head can broadcast the information to other cluster in the network due to which it achieves scalability in the networks.
When the cluster is created all the nodes in the network uses beacon in a given time frame. If the nodes get the response within the defined time frame then that node joins the cluster. If the node receives a packet it checks whether the destination is in its own cluster, if it is then it sends the packet to the nearest node towards the destination. And if the cluster head receives the request it verifies that the destination node lies in different cluster or in its own cluster. If the destination is in its own cluster then cluster head sends LREP in response of LREQ initiated by the source node. And if the destination is not in its own cluster, the cluster head sends the packet to its neighbor cluster. This protocol has some limitations since network is relying upon cluster head. If the cluster head is compromised, it can result whole network to become fail. Also excessive broadcasting is used in this protocol which can affect bandwidth badly.
ROVER Robust Vehicular Routing  routing protocol is based on Geo Cast and the main purpose of this protocol is transmit the information to every node in the network in a specified Zone of Relevance (ZOR). Broadcasting packets in this protocol are the control packets whereas unicasting packets in this protocol are data packets. Due to the usage of broadcasting control packet in the network, this protocol can cause overhead. Latency is high. Excessive use of broadcasting can affect bandwidth badly.