This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Mobile Ad-hoc networks are characterized by constant topology changes, in the absence of fixed infrastructure and a centralized control. Traditional routing algorithms prove to be incompetent in such a changeable and unstable environment. Ad-hoc routing protocols such as Dynamic Source Routing (DSR), Ad-hoc On-Demand Distance Vector Routing (AODV) and Destination sequence Vector (DSDV) have been anticipated to resolve the multi hop routing problem in Ad-hoc networks. Broad research has been done in comparing the different projected ad-hoc routing protocols under varying network scenarios. Routing overhead, packet delivery ratio, end-to-end delay, path optimality, and throughput are some parameters commonly used in the comparisons .In this survey paper we had analyzed the performance of the routing algorithms under various networks conditions.
Keywords- MANET, DSR. AODV, DSDV, VBR,CBR
Mobile ad-hoc network (MANET) is a group of wireless mobile nodes which dynamically forms a transitory network without the use of any preexisting network infrastructure or centralized supervision. Routing protocols in ad-hoc networks must have the property to automatically adjust to environments that can diverge between the extremes of mobility and bandwidth requirements The performance of MANET is related to the competence and effectiveness of the routing protocols in adapting to repeatedly changing network topology and link status.
2. AN OVERVIEW OF PROTOCOLS
The MANET routing protocols can be classified
in three categories:
Table driven (proactive) protocols:
The needed information for routing purposes is stored in the tables; tables are frequently updated through control packets sent by other nodes in the network. The updates can also respond to topological transformation of the network. Example is DSDV.
On-demand (reactive) protocols:
In contrast to table driven routing protocols, calculate the path to a specific target only when required, so a routing table does not required to be update all the nodes as entries in each node. When a source wants to send packet to a target, it call up a route discovery mechanism to find the route to the target. The route remains valid till the target is available or until the route is no longer desirable.
Examples are AODV, DSR.
Hybrid Routing Protocols
Based on properties of both table and demand driven routing protocols, some hybrid routing protocols are anticipated to combine advantage of both proactive and reactive protocols. The most typical hybrid one is zone routing protocol.
Fig. 1. Categorization of ad hoc routing protocol
The following subsections briefly describe the three ad hoc routing protocols which are considered in this paper. The first one (DSDV) is fully table driven whereas the second one (DSR) is fully on-demand based. The third one (AODV) has adopted some of the characteristics of the both the categories.
A. Destination Sequence Distance Vector (DSDV)
DSDV  is a distance vector routing protocol, based on the Bellman-Ford routing algorithm. It is a proactive routing protocol and a hop-by-hop protocol in which every node in the network maintains a routing table. A routing table has all of the possible destinations nodes and the number of hops to each destination. Each entry in the routing table has a sequence number assigned by the destination node, implies the freshness of that route, thereby avoiding the formation of routing loops. By periodically update messages, routing tables maintain consistence.
B. Dynamic Source Routing (DSR)
DSR  is a source routing protocols, and requires the sender to know the end to end route to destination. It is based on two main processes: (a) the route discovery process which is based on flooding and is used to dynamically discover new routes, maintain them in nodes cache, (b) the route maintenance process, periodically detects and notifies networks topology changes. Discovered routes will be cashed in the relative nodes .DSR is a fully reactive routing protocol
C. Ad Hoc on Demand Distance Vector (AODV)
AODV  combines some properties of both DSR and DSDV. It uses route discovery process to cope with routes on demand basis. However, it adopts DSDV like hop-by-hop routing tables for maintaining routing information. Hence AODV is a reactive protocol; it doesnâ€™t need to maintain routes to nodes that are not communicating.
3. SURVEY OF PERFORMANCE OF AODV, DSR AND DSDV UNDER VARIOURS NETWORK CONDITIONS
3.1. Performance Comparison when streaming MPEG4 traffic
Performance studies of these routing protocols have assumed Constant Bit Rate (CBR) traffic. Real-lime multimedia traffic generated by video on demand and teleconferencing services are mostly Variable Bit Rate (VBR) traffic. Most of these Multimedia traffic is encoded using the MPEG standard. (IS0 Moving Picture Expert Group). When video Traffic is transferred over MANETs a series of performance issues arise. Simulation studies have been done  and it shows that DSDV performs better than AODV and DSR. However all three protocols fail to provide good performance in large, highly mobile network environments all the three protocols f
a.MPEG4 traffic modeling
MPEG-4 is an ISO/IEC standard developed by Moving Picture Experts Group, the committee that also developed the Emmy Award Winning standards known as MPEG-1 and MPEG-2. MPEG4 offer high excellence video at a relatively low bit rate and has the potential of real-time adaptive encoding and has become popular standard for, various network applications, especially the Internet . A range of techniques such as Markov chain models, Autoregressive Moving Average (ARMA) models, Bernoulli process models, spectral characterization and transform-expand-sample (TES) model have been used to model network traffic. We studied the behaviour of three routing protocol when MPEG4 traffic is modeled using the TES modeling process 
Various experimental results had been studied and the one of performance metric that is used to determine the performance is packet delivery ratio that is the number of packets that was delivered to the destinations as a fraction of the sent packets.
Simulation study  shows that when the maximum speed at which the nodes moved in the environment was varied, it is not feasible to stream MPEG4 traffic in highly mobile ad-hoc networks .Experiments shows that Overall DSDV showed better performance than AODV or DSR. But the difference in performance was within 2%.
Another study has been done when the rate factor of the MPEG4 stream was varied, results shows that if the rate factor increases the packet delivery ratio decreases. Rate factor scales up (down) the MPEG4 stream. More packets are required to deliver the MPEG4 stream to the target if the rate factor is higher; as a result the load on the channel is increased. The performance got degraded when the rate factor increases, experiments shows that the channel cannot take the load even at the slow speed of 10 m/s, therefore the data packets are dropped. This shows that there is a limit to which the MPEG4 stream can be scaled up when it is streamed in mobile ad-hoc networks.
The experimental study  shows that various rate factors confirms dissimilar performance .If rate factor is between 0.5 to 1 it shows good performance about 90%, but further increase 1 to 1.5 % performance got degraded by around 15%, this is to be continue around rate factor 3, after which the performance keep on dropping but the performance drop will be at slower rate.
This study shows the performance of three ad-hoc protocols (AODV, DSDV and DSR), when MPEG4 traffic is streamed through the network. Study shows that the number of nodes, speed and the rate factor has a great impact on the performance of the three protocols. Overall DSDV showed better performance than AODV or DSR. But the difference in performance was within 2%.
3.2. Performance analysis in realistic environment
The motivation of this study  was due to experimentations where it found that due to the use of simple radio propagation model that did not take into account transient links caused by small-scale fading, AODV and DSDVâ€™s performance did not match simulation. Simulation studies explain each protocolâ€™s key conduct that direct to the following conclusions:
(i)Routing through steady routes has a great significance, (ii) Hop count is the source of poor performance, (iii) for the simulation of ad-hoc protocols in indoor environments, the 2-ray ground radio propagation is not appropriate and (iv) local recovery is important. In this study it has been observed that in addition to mobility, channel fading plays an important role in determining protocol performance. DSR has its aggressive nature in discovering routes and it uses the cached routes for the retransmission of packets when active routes failed thus DSR shows the best performance in comparison to AODV at the expense of increased overheads. AODV, however, required the source to perform route founding on every occasion when link(s) on the established path failed. An experimental study shows that AODV required a considerable amount of time to establish a dependable and consistent route. When more than a few hops were required, AODV had difficulty in route establishment. On the other hand, DSDV was not as significantly affected by transient links when compared to DSR and AODV.
3.3. Performance analysis under security attacks
Mobile Ad-hoc Networks has unique characteristics such as shared wireless medium, no fixed infrastructure lack of centralized control and open peer-to-peer network architecture, and due to these characteristics MANET poses various security issues and challenges. This study aims to compare three routing protocols, DSDV, DSR, and AODV under security attacks, when the presence of misbehaving nodes have been investigated. Network performance is estimated in terms of routing overhead, average packet delay and normalized throughput when a fraction of nodes misbehave. Simulation results  show that even though the performance of all three routing protocols humiliate, DSDV is the most robust routing protocol under security attacks.
The study of Misbehavior of nodes has been used to differentiate networks that are under security threats. We studied two types of misbehaviour: a selfish behaviour and a malicious behaviour . Selfish nodes are those nodes that do not cooperate; main intension is to save the battery life for their own communications, but do not intended to directly damage other nodes. All together we define three routing behaviors of nodes.
1) Type 0 well-behaved nodes: Nodes behave adequately according to a routing protocol including route discovery, maintenance, and packet forwarding and receiving.
2) Type 1 selfish node: In this model, every packet sent to the selfish node is dropped by it; it does not perform the packet forwarding. Thus, it immobilize the packet forwarding task for all packets that have a source address or a target address different from the current selfish node address.
3) Type 2 selfish node: In this model, the node does nothing with the packet sent to it; thereby no execution function is performed. The selfish node can be considered as a relax node within the network, as it discontinue to contribute to the network maintenance.
In general, DSR and AODV undergo a lot from the two types of selfish nodes, while DSDV shows stable performance under security threats though its performance also degrades. To compare the protocols, network performance is evaluated according to the following parameters:
a. Normalized Throughput: In a low mobility network and in the absence of selfish node in the network, AODV and DSR reach higher throughput (i.e., deliver approximately 80% of the offered load) than DSDV (which delivers just about 60% of the offered load). The throughput of all protocols turns down when the fraction of selfish nodes raises. On the other hand, the normalized throughput of AODV and DSR drops quickly in the presence of Selfish node; for AODV from about 70% to about 10%, and for DSR from about 60% to about 10%.This is because the selfish nodes do not cooperate well as well-behaved nodes in the network.
b. Packet Delay: Simulation study shows that DSR always has a higher average packet delay among the three and DSDV the lowest with or without selfish nodes, because DSR desires time to discover a path on demand of the source, or when the link break happens; while DSDV is a proactive routing protocol and finds route from time to time. AODV results in a modest packet delay in the middle.
c. Routing Overhead: As the normalized throughput of DSR and AODV requires less routing overhead because normalized throughput of these two protocols suffers from the selfish nodes, so more packets are dropped in the network. In contrast, DSDV has a roughly stable overhead despite the survival of selfish nodes and there count. This is because DSDV is table-driven with fairly unwavering routing control overhead.
d. Normalized Routing Load: AODV normally has the maximum normalized routing load among the three protocols and DSDV the least with or without selfish nodes. This shows that DSDV is the best in finding routes optimal to the shortest paths.
On comparing three routing protocols, DSDV, DSR, and AODV under security attack, when Network performance is evaluated in terms of normalized throughput, average packet delay, routing overhead and normalized routing load, and when a percentage of nodes behave selfishly, simulation results shows that although the performance of all three routing protocols degrades, DSDV is the most vigorous routing protocol under security attack. This shows that a proactive routing protocol has the potential of not including misbehaving nodes in advance and sinking the impact of security attacks.
3.4. Performance analysis of resiliency under the range attack
The range attack does not require a node to be compromised. An opponent only has to get close enough to a node to soothe or amplify radio signal. In the range attack the transmission range of a wireless node is frequently customized to cause a changeable network topology. An enemy merely needs to get close to a node to attenuate or amplify the radio signal.
According to the simulation results , DSDV endow with the best resiliency against the range attack when the primary application requires a very short transmission delay less or equal to 0.1 seconds. AODV endow with the best resiliency against the range attack, when the primary application tolerates delays up to 2 seconds, AODV offer the most excellent resiliency against the range attack.
When short transmission delay is required.
DSDV offers the best resiliency against the range attack when short transmission delay is required. Here, AODV provides the worst overall resiliency against the range attack.
2. When long transmission delay is allowed. When an application allows a comparatively long transmission delay, AODV is the most resilient against the range attack. Reason for the good performance of AODV is that it is more swift in locating a new route than DSDV.
In the range attack an rival periodically soothe the power of transmitted radio signal of a wireless node. This causes periodical topology changes. The simulation results indicate that the selection of a defense mechanism is situation dependent. If an ad hoc network is mainly used by an application have need of a transmission delay less or equal to 0.1 seconds, DSDV provided the best and AODV the lowest resiliency against the range attack.
3.5. Performance analysis under network load deviation
In this study  we had the performance comparisons of the DSDV, AODV and DSR routing protocols by means of modified path optimality metric that we name it weighted path optimality, network's load deviation that have an effect on nodes and can be a metric for load balancing and average end-to-end delay.
a. Weighted path optimality: DSDV make use of Bellman-Ford algorithm to find the shortest path between source node and target node. Therefore DSDV act upon on the whole well. In terms of performance, DSR and AODV come next respectively. The closer the value to zero the better weighted path optimality. In DSR, any node within a path, if to be learned of a shorter route to the target will update this recent route. But in AODV these information will be received merely by neighbors of the source node. Hence DSR performs better than AODV.
b. Average end-to-end delay: DSDV has the shortest average end-to-end delay. The benefit of these protocols is that a route to a target is straight away obtainable, so no delay for route discovery is experienced when there is a request to send a packet. In addition DSDV go for shortest paths by make use of Bellman-Ford algorithm, where as reactive protocols locate longer route than DSDV and may be need to determine paths in the beginning of transferring.
c. Deviation of network's load: Increasing paused time leads to fewer networksâ€™ topology changes and so the selected paths by the protocols will be more steady and firm. Therefore, protocols have more tendencies to use meticulous nodes (through these paths). This, in turn, leads to supplementary loads on these nodes and as a result more deviation of network's load. DSDV is a proactive protocol and focuses on the shortest paths which are calculated from routing tables recorded in nodes and make use of only those nodes which are in the course of these paths. Therefore DSDV has the most unpleasant performance and DSR has the finest performance.
4. COMPARISION OF ROUTING ALGORITHM BASED ON SURVEY
Table: 1 Comparison of DSDV, DSR and AODV
When Streaming MPEG4 Traffic
Better than DSR
In Realistic Environment
Not effected by transient links
Degraded as compare to DSR
Better than DSR
Average packet Delay with selfish nodes
Lowest average packet delay
Higher average packet delay
Resiliency Under Range attack with Low transmission delay
Resiliency Under Range attack with high
Lower than AODV
Varies sometime as good as AODV
Better than AODV
Next to DSR
Average end- to- end delay
In this survey paper the analysis has been done for three routing protocol AODV, DSR and DSDV under various network conditions. DSDV shows the best performance under security attacks while AODV shows the worst performance, while streaming the MPEG4 Traffic DSDV shows the best performance and DSR shows the worst performance. Thus the performance of these routing protocols is application specific and shows different performance under different network environment.