Performance Analysis of QoS Parameters of MANET on Mobility and Energy based Model with Different MANET Routing Protocols

1Department of Computer Science and Engineering, Siksha ‘O’ Anusandhan University, Bhubaneswar 751030, Odisha, India; ivijaya@rediffmail.com 2Department of Computer Science and Engineering and IT, V.S.S.U.T, Burla 768018, Odisha, India; amiyaamiya@rediffmail.com 3Department of Computer Science and Engineering Retired Professor, IGIT, Sarang 759146, Odisha, India; bhagabat.puthal@gmail.com 4Department of Computer Science and Engineering, Siksha ‘O’ Anusandhan University, Bhubaneswar 751030, Odisha, India; mishradebahuti@gmail.com 5Department of Computer Science and Engineering, Siksha ‘O’ Anusandhan University, Bhubaneswar 751030, Odisha, India; sandeepkumar04@gmail.com


Introduction
Mobile Ad-hoc Network (MANETs) in recent years has gained attention of researchers. Research has been pro-gressed in diverse aspects of MANET 1-9 . The motivation behind the various diversities is due to its characteristics such as; dynamic topology, bandwidth constrained and adhoc network (MANET) -ANMAS (Adhoc network multicasting with Ant system). In this method, ants 'pheromone' was used for the indirect communication to obtain dynamic topology change information and achieve a 'tolerable' multicasting group. Author 21 measured the inefficiency of the overlay multicasting. Author 22 proposed and presented results of a shared tree variant of Shared-Tree Multicast Zone Routing Protocol.
Each type of application that a MANET supports can be identified by evaluating the performance of mobile adhoc wireless networks 23,24 . The network performance metrics used, analyzed and evaluated for our study are network layer parameters as listed 25 a) TP =throughput b) PDR = packet delivery ratio c) AD=average end-to-end delay and d) NRO= normalized routing overhead. In our work, we considered a proactive and two on-demand reactive routing protocols i.e.; DSDV and (AODV, DSR). The mobiles devices in the network get connected only when there is a demand for it. The algorithm used in ADODV and DSR is gateway discovery.

MANET Routing Protocols
In adhoc networks routing is a complex problem due to its dynamic nature. The process of designing communications and the right protocol for the type of network is quite a challenging process. Considering the most important aspect of communication is to establish a route, maintain and transfer data between nodes or devices, protocols play an important role 11 . Numerous studies and researches have been done in the area of designing a protocol 1-3 . Depending on the type of complex network under study, researchers have suggested multitude of protocols 12 . To evaluate the efficiency of a protocol, mobility and traffic are two the important aspects that should be considered. These protocols support a broader range of MANET applications. The best-effort basis is chosen to establish and maintain routes between nodes 13 .

Strategies of Routing Protocols
The network structures supporting any routing protocol are classified into flat, hierarchical and geographical I. Vijaya, Amiya Kumar Rath, Bhagabat Puthal, Debahuti Mishra and S. Satapathy position assisted. There are three very common and predominantly used routing strategies; proactive strategy, reactive strategy and hybrid strategy [24][25][26][27][28][29] . Proactive strategies are table-driven and reactive strategies are demand-driven. Hybrid is a combination of proactive and reactive strategies. Classification of adhoc protocols at a broader level is shown in Figure 1. The explanations along with their characteristics of these protocols are briefly explained in the following sub-sections.

Proactive Strategy
In Proactive strategy, each node in the network maintains the routing information. Maintaining the route of the node helps to forward the packet to the desired node making search faster. Examples of these protocols are Optimized Link State Routing (OLSR), Destination Sequenced Distance Vector (DSDV), Fisheye State Routing (FSR), and Fuzzy Sighted Link State (FSLS). The limitation of this category of protocols is that they utilize high network resources due to continuous updation of routing table information. Table 1 describes the characteristics of the proactive type of protocols 27,28 .

Reactive Strategy
The basic principle of this scheme is that each node in the network maintains active route destinations only. This scheme overcomes high resource utilization as data sending and receiving is not continuous. Communication • A routing node is maintained for each node with the destination information, next hop and required no. of hops to reach the required destination • Periodically route information is updated based on the route information broadcasted.
• Two packets are carried to the nodes, one containing routing information (full dump) and second carry only the delta changes from the full dump (incremental). • Sequence numbers are maintained for the routes and updates to the routes are based on the sequence number. • Routing table is updated ignoring the old sequence number with the recent sequence number • Already existing sequence number if matches, then information of next hop and # of hops are considered for update (best metrics). • The new information is then broadcasted into the packet.

WRP (Wireless Routing Protocol)
• It is modified version of DSDV.
• A set of four tables -distance, routing , link cost and message retransmission are maintained.

OLSR (Optimized Link State Routing Protocol)
• Each node uses its most recent information to route a packet.
• Multipoint Relay nodes are selected based on the greedy algorithm.
• OLSR protocol performs hop by hop routing.
• The source node communicates with its two-hop neighbors through multi point relay nodes.

FSR (Fisheye State Routing)
• The network structure is like 'fisheye.
• Each node has only routing information of the nearest node.
• The information contained is exact and accurate.
Performance Analysis of QoS Parameters of MANET on Mobility and Energy based Model with Different MANET Routing Protocols overhead is also reduced and is mostly used. Table 2 describes the characteristics of the reactive type of protocols [27][28][29][30] . • Each node establishes routes only when requested by the source node and caches it. • Nodes maintain a tree like structure containing the local connectivity of information and when the cache is unable to serve the requested data, source flags a RREQ route request broadcast packet.

•
The receiving node checks the availability of the information of the destination node in its cache and sends a (RREP) route reply.
• Nodes keep track of the RREQ's source IP address and broadcast ID.
• The nodes that are part of an active route contain connectivity information by broadcasting local 'Hello' messages to its neighbors. • The source node reinitiates route discovery as soon as it receives route error (RERR) indicating link break to the destination.

DSR (Dynamic Source
Routing) • The sender node correlates the chain of nodes to reach the destination for transmitting the packets. • For transmission of packets, the source node searches route cache for destination. • Each packet contains sequence of hops taken by the route request packet during route discovery.

HSR (Hierarchical State
Routing) • The nodes are partitioned into clusters and each cluster has a cluster head that is responsible for passing its information to neighbor nodes. • The structure used is hierarchical structure which combines clustering technique with the location management. • The cluster head is used as gateway to deliver the data to any part of the other network.

TORA (Temporally Ordered
Routing Algorithm) • The source node determines the direction of the link between two nodes and uses a parameter 'height' to establish a link for the shortest path. • The source initiates a route to destination by broadcast and the node that has the destination replies with an update of the parameter height. • The links from source to destination are directed links in descending order of height. • The invalid links are removed from the network when the node cannot find the neighboring links.
• It combines two processes of route establishment and route maintenance. • It used the link reversal algorithm.
• The source establishes a route by query packets and failures are recorded to maintain the route

Hybrid Strategy
The basic principle of this scheme is that both the proactive and reactive strategies are clubbed on the network. The network is further divided into clusters (intra & inter clusters). In the intra-clusters the proactive protocol is used and reactive protocol is used in inter-cluster. Table  3 describes the characteristics of the hybrid type of protocols 31,32 . The nodes are divided into zones and each zone has a set of interior nodes (each node having information of its neighbor node). The routing in intra-zone is done by proactive approach.
The neighbor discovery is done by neighbor discovery protocol. The outside zone route discovery is done by reactive approach The source nodes initiate the route discovery, through a sequence of checks i.e.; check in intrazone, then check in the peripheral node and finally check outside the zone. In this way route discovery is been done. The nodes are grouped into zones and nodes within the proactive zone maintain information with respect to the central node. The zone radius determines the nodes belonging to a particular zone. In this scheme, if a node is unable to retrieve information of destination node it uses reactive mechanism of (REQ-REP) to establish the route The proactive zones are the collectors of packets.
The parametric comparison 27 of above mentioned three routing protocols along with their relative advantages and disadvantages 28 are shown in Table 4 and Table  5 respectively.

MANET Mobility Model
Mobility is defined as the movement of mobile users in a network. Theoretically mobility model can be used to represent these mobile users. The mobility model uses protocols to set up routes between nodes and also help in discovering of routes 33 . To represent a model of the mobile devices in the network, analytical or simulation models are used. The relationship between the mobility model and the protocol performance is represented by the Figure 2. The mobility models play an imperative part in assessing the ad hoc protocol performance. It describes multitude of operations like movement pattern and presence of mobile nodes in the location, the speed of a node and the acceleration of nodes with the change of the parameter time 29 . The type of model used influences two aspects i.e.; performance and derivation of solution to use the best model for complex situation. The most widely and predominantly used mobility model in MANET simulations is the random waypoint model that is described in following section 34 .

Random Waypoint Model
Random way point is the simplest and widest used model in most of the simulations was first proposed by Johnson and Maltz 35 which became a benchmark mobility model to evaluate the MANET routing protocols. It was available with the widely used network simulator NS2. The basic principle of operation of this model is the nodes move independently to a randomly chosen location (also termed as destination) with a constant velocity chosen uniformly. The velocity and direction are independent of each other. A pause time is included between the change in destination and velocity of a node. At the destination, the node stops for the pause period defined. After the expiry of the pause time, another random node is cho-sen as destination within the simulation area. The speed is also chosen randomly during the new cycle of simulation 35 . This process is repeated until end of the cycle. Several researchers have proposed a modification to this by setting the velocity to non-zero minimum speed. With this the network becomes more stable and reliable, hence converges simulation results to a constant and stable level 36,37 . We adopted this improve model in our simulations.

Energy Model for MANET
In the earlier sections, we stated the primary goal of a routing protocol is efficient when a route is established between the source and the target destination. The availability of link and route to a destination node 38 determines the survivability of the network. To establish a route to the destination without affecting the performance, energy should be conserved for critical nodes 40 . Ultimate goal is to conserve energy of the nodes 38,39 . In our model, energy awareness is implemented as a part of routing protocol at the network layer. When the number of active connections is more, we use reactive routing protocol like AODV. On the contrary DSR saves bandwidth and reduces power consumption as it doesn't use periodic routing relay. During link failures also, source nodes have the capability of checking its own cache for an alternate route. At any point of time the activities performed by the nodes is either transmit data packets, or receive data from neighbor nodes or idle neither transmits nor receives. The energy consumption is directly proportional to the bandwidth and size of the packet being transmitted.

Simulation Model for MANET
Our simulation is performed on two different models to evaluate MANET routing protocols. First is mobility model and second is energy model. The important characteristics of mobile nodes are the node positions and speed. The node positions are assumed to change constantly with respect to time. The pause time of a network decides the time taken for the nodes to change their positions or start sending packets to destination. Hence, in our work to observe the performance of a network variable pause times are considered. Also speed of nodes is another variable considered for evaluation of network performance. Similarly, in energy model the power of mobile nodes along with their mobility are considered as the essential properties of mobile nodes. Like mobil-ity model, to evaluate the performance of network, we considered variable pause time and speed of nodes. In the energy model we considered some extra parameters as compared to mobility model such as initial energy, idle, or receiving or transmission or sleep power etc. The working procedure proposed work has been outlined in Figure 3.

Performance Parameters
Performance parameters are used to measure the efficiency of a routing protocol in different situations. In MANET, the mobile nodes are not constantly present at a single position, so it is a very challenging task to maintain the network stability along with its performance. Two different models are taken to simulate a MANET. One is mobility model and the other is energy model. In mobility model the node positions and speed are considered as the important characteristics of mobile nodes. Similarly, in energy model the power of mobile nodes along with their mobility are considered as the main properties of mobile nodes. As we know, mobile nodes move constantly and their performance depends on the power consumed by the mobile nodes hence a simulation has been performed with energy model for different routing protocols. To evaluate the performance of MANET for two different models we have considered four parameters as discussed in section 1 such as; TP, PDR, AD and NOR. Along with these four parameters one extra parameter has been taken for energy model that is average energy consumption (AEC).The parameters are explained mathematically as follows from equation (1) to (4). (1)

Simulation Results and Discussion
This section demonstrates about the results obtained from ns2 simulation of MANET using routing protocols AODV, DSDV and DSR. Table 6 shows the simulation environment setup for MANET using a mobility based model for different pause times. Here, the pause times are considered as 0, 30, 60, 90, 120, 150 seconds etc. All simulations are performed using NS2. 35. The graphical representation of networks is visualized using NAM editor available in NS2. Similarly, the Table 9 shows the simulation environment setup for MANET using mobility based model for different speed of nodes. Speeds are considered like 10, 30, 50, 70, 90 m/sec etc. Another model for MANET has been simulated for the energy consumption by the node with respect to movement of nodes. Power consumption is considered as one of the important feature of a MANET which increases or decreases the efficiency and stability of the network. Energy consumption has been evaluated by considering different pause time and speed of the nodes as discussed earlier in case of mobility based model.

Simulation results for mobility based model
The simulation environment setup for mobility based model with varying pause time is given in Table 6, the comparison of pause time with respect to TP and PDR, AD and NRO has been shown in Table 7 with Figure 4 and Table 8 with Figure 5 respectively. In all the cases the numbers of nodes are fixed to 50 and the pause time of mobile nodes is varying starting from 0 to total simulation time with a difference of 30 seconds. From Figure 4 it can be clearly understood that the reactive routing protocol DSR dominates its peers that is AODV and DSDV in terms of throughput and. It can be observed that for DSDV the throughput is very poor at 0 pause time but increases as pause time increases. Similarly in terms of delay and routing overhead DSR performs better as compared to others.
Similarly, simulation environment setup for mobility based model with varying speed is given in Table 9, the comparison of speed with respect to TP and PDR, AD and NRO has been shown in Table 10 with Figure 6 and   Table 11 with Figure 7 respectively. In this case, speed of the nodes was varied from 10 m/s to 90 m/s where as the number of nodes and pause time is kept as fixed value.
We proved that the protocol DSR outperformed for throughput, packet delivery, delay and routing overhead as compared to AODV and DSDV. Also for DSDV the throughput increases significantly as the speed of nodes increase. It can also be observed that for AODV protocol the routing overhead is high as compared to DSR and DSDV with respect to varying speed of nodes in mobility model.

Simulation results for energy based model
The simulation environment setup for energy based model with varying pause time is given in Table 12, the comparison of pause time with respect to TP and PDR, AD and NRO and AEC has been shown in Table 13 with Figure 8, Table 14 with Figure 9 and Table 15 with Figure 10respectively. There is a similar type of comparisons made for evaluation of routing protocols based on energy model like we have already done for mobility based model. The node while moving in the simulated space and transmitting the packets, the average energy consumption by the nodes is calculated. From the graphs we observe that in the energy model, DSDV protocol out performs than others. There is no significant improvement in TP and PD as compared toDSR. DSR protocol still would be considered for better TP& PD. Most importantly it can be clearly seen from the graphs the average energy consumed by DSR protocol significantly low as compared to DSDV and AODV. This means we can achieve high performance from a MANET with DSR routing protocol with less energy consumed by the mobile nodes. All these observations have been made by varying the pause time.
Similarly, simulation environment setup for mobility based model with varying speed is given in Table 16, the comparison of speed with respect to TP and PDR, AD and NRO and AEC has been shown in Table 17 with Figure  11, Table 18 with Figure 12 and Table 19 with Figure  13respectively. Similarly a different set of observations have been made to evaluate performance of network with varying speed. We can observe the same behavior as previous. But the most important alteration we can see there is an increase in routing overhead for DSR protocol as the speed of nodes increases. However the energy consumption is still low as compared to others.

Conclusion
The experimental evaluation of the MANET routing protocols AODV, DSDV and DSR on mobility based and energy based models considering the QoS parameters throughput, packet delivery ratio, and average delay and normalized routing overhead resulted in drawing the aforesaid conclusion. We can conclude that the DSR protocol dominates all other protocols like AODV and DSDV. The Dynamic Source Routing protocol in mobility and energy based model for throughput, packet delivery ratio performs well than AODV and DSDV. The adverse result is that, with the increase of node speed, routing overhead increased for DSR. Positive aspect of DSR was that average energy consumption was quite low in contrast to AODV and DSDV.