An agent based routing search methodology for improving QoS in MANET

,


INTRODUCTION
MANET is a communication paradigm, which does not require a fixed infrastructure; they rely on wireless terminals for routing and transport services [1]. Nodes rely on each other to keep the network connected and to move information. The act of moving information from source node to destination node is called Routing. The routing concept basically involves two steps [2][3]. First, determine the optimal routing path and then transfer the information packets through the network. Routing protocols use several metrics to calculate the best path for routing the packet to its destination [4][5][6]. These metrics are a standard measurement that could be, for example, number of hops, which is used by a routing algorithm to determine the optimal path for the packet to its destination. Routing is mainly classified into static routing and dynamic routing. Static routing maintains a routing table [7][8][9]. Dynamic routing refers to the routing strategy that is being learned by an interior and exterior routing protocol.

PROBLEM STATEMENT
The problem with AODV is its single path selection, with a reactive approach that claims to be best at a particular instant of time, but topology is dynamic. Hence it may be changed and the network may have a better route but AODV does not check for it and continue, to the already selected path another scenario can be if any node moves away from the established path, between source to destination. Then route error comes and source has to broadcast the route request to its neighbor hence re-route discovery that increases control overhead, network latency and decreases the overall throughput of the network. The occurrence of re-routing nodes could disrupt the routing operation in MANETs. To overcome this behavior, the integrity of the network nodes should be considered in the route selection process combined with the hop count. That is the reason the concept of mobile agents in AODV is used, also called AB-AODV.

METHODOLOGY
In the proposed algorithm, any path from the source node to the destination node is a feasible solution. The optimal solution is the shortest one. At the beginning a random population of paths is generated which represents feasible or unfeasible solutions. Unfeasible solutions are paths that do not end at the destination. A prey hunting corresponds to any possible solution to the optimization problem. Thus each prey represents a path which in tum consists of a sequence of positive integers that represent the IDs of nodes through which a routing path passes with the source node followed by intermediate nodes (via nodes), and the last node indicating the destination, which is the goal. The total grey wolf count is equal to the number of nodes.

PROPOSED METHODOLOGY
This agent-based Ad Hoc On-demand Distance vector routing (AB-AODV) is an improved AODV routing protocol based on mobile agents. In AB-AODV mobile agents work on the method of indirect coordination between agents, based on the footprint left in the environment by a subsequent action. Mobile agent updates routing information as well as the routing table information left by other mobile agents. Using this information, mobile agents are able to get the routing information about the mobile nodes it has not visited yet. Each agent is considered free to all other agents in the context of its code and data. An agent can move freely in the entire network. Due to the small size of the mobile agent, transmission overhead is very low. Agents communicate with other agents indirectly by mobile routing table in order to _nd out the shortest best route. As an agent helps to _nd out the best possible route between mobile nodes so if any time any agent moves to another location. Whenever a node is out of the range of other nodes then through an agent, a node can switch from the current node to the better route node. Therefore route latency can be minimized by an agent because agents save the time that consumed in the discovery of route again else in AODV whenever RERR (route error) comes it stars from the initial stage and again broadcast the RREQ (route request) packet to its one-hop neighbour nodes and repeats the whole route discovery process. By agent, AB-AODV avoids the unnecessary route discovery process by searching the alternative paths and updating these paths into the routing table of mobile nodes. Agents not only minimize the routing overhead and network delay but also improve the overall throughput of the network.

MOBILE AGENT
Mobile Agents are simple packets that carry the data and search into the network for the available routes. Agents leave the knowledge acquired through study of the network behind the nodes that they travel to, so that other agents can use this information. As in computer networks there are various devices that are connected to the networks where mobile devices. Mobile Agents help to accumulate and distribute connectivity information for a dynamic wireless network. Topology of the mobile network is dynamic in nature because the connection between mobile nodes is established and destroyed when nodes move in the ad hoc manner. Sometimes nodes are in and out of range of each other. The mobile agent stores the past movement where it has been. When a node receives information from a mobile agent and updates its routing table with best available routes. The routing agents communicate indirectly to each other by writing the information to the routing table of mobile nodes but do not read information from the mobile nodes. When an agent comes to such a node it updates the node's routing information by its(agent) collective information as well as it reads routing information from the local routing table that builds as per the routing information collected by other agents. Hence by the help of the local routing table of each mobile node, agents can communicate with each other indirectly and help to determine the best possible routes to the mobile node.

DETERMINATION OF AVERAGE NUMBER OF MANET
In MANET that consists N number of mobile nodes, if a node is able to create a new mobile agent after every S seconds with a determinate lifetime of T seconds, then after a relatively long period in comparison to the agents lifetime, the average number of agents (A) is where, A = average number of mobile agent N = number of mobile nodes T E = expiry time of a mobile agent S = time interval to create a new agent

EVALUATION OF FITNESS FUNCTION
The fitness function F(x) is defined as follows: where, NO = Normalized Overhead AD = Average End to End Delay PD = Number of Packet drop PDR = Packet Delivery Ratio k = Proportionality constant used for the optimization of_tness function. Value of k lies between 0 & 1, i.e. 0 ≤ k ≤ 1.

Simulation Parameters
Simulations have been carried out in order to evaluate routing protocol. We focused our attention on the evaluation of network performance in terms of Drop Packet, Packet Delivery Ratio and All Delay, Average End to End Delay and Throughput of a mobile ad hoc network where a number of nodes are varying. Simulation environment to calculate the real time performance of the aforesaid quality of services parameters in Table 1.

Simulation Scenarios
Two simulation scenarios are used to simulate the effect of rushing attack and the effectiveness of prevention techniques in mobile ad hoc networks.

Implementation of Traditional AODV Routing:
In this simulation scenario, first of all we construct a mobile ad hoc network and configure it with the help of AODV routing protocol. Then an illegal node is set over MANET and gains the performance of AODV protocol with the help of trace file.

Implementation of AB-AODV Technique:
In this simulation scenario a MANET is constructed and configures it with the help of AODV routing protocol after it an illegal node is constructed over MANET and obtained the performance of AB-AODV protocol. Finally simulate the effect of prevention techniques in the form of PDR, Throughput, Endto-End Delay, Drop packet and Lost Packet etc.

PERFORMANCE PARAMETERS
The following five performance parameters are used to study the protocols efficiency, adaptability, and scalability: 1. Throughput: It is defined as the total no. of number of packets received per second over simulation time.

Average End-to-End Delay:
It is the average time delay incurred from the time when a data packet is sent from its source node until the data arrival at its destination node, divided by total number of data packets delivered at destination. 3. Packet Delivery Ratio: It is the ratio between the data packets delivered to the destination and those generated by Constant Bit Rate (CBR) sources. This evaluates the ability of the protocol to discover routes and its efficiency.  Figure 2 shows comparison of network drop data packet with AODV and AB-AODV.

Drop Packet:
Drop data packet in presence of AODV is low and high after applied proposed technique.

Packet Delivery Ratio (PDR):
The Table 2 shows the PDR for the AODV and AB-AODV. There is improvement in the Packet Delivery Ratio in the proposed technique. From the table it is visible that the PDR for AODV is less than the Packet Delivery Ratio for Proposed Method for nodes in the MANET.

End-To-End Delay (E2E):
The Table 3 shows the E2E delay for the AODV and proposed technique. E2E Delay is decreased In Proposed Technique compared to AODV. The comparative End-To-End Delay graph is shown for the AODV and Proposed method. From Table 3 it is visible that the End-To-End Delay for Proposed Method is less than the Packet Delivery Ratio for AODV for nodes in the Mobile ad hoc network.

Throughput:
The Table 4 shows the throughput for the AODV and proposed technique. Throughput   is increased in proposed techniques compared to AODV. The comparative analysis of the AODV and Proposed method represent that the throughput for AB-AODV is less than the throughput for AODV for nodes in the Mobile ad hoc network.

CONCLUSION
The improvement of routing protocol AODV with the help of mobile agents is proposed in this paper. Simulation is done under Network Simulator named NS-3 and observed results that AB-AODV is better than AODV in the aspects of throughput, delay and packet delivery ratio. We did analysis on 5, 10, 15,20 and 25 nodes in all the cases AB-AODV proved better than Traditional AODV.

FUTURE WORK
A new protocol should be automatic to adapt to the changes in topology and data trac. So, in future by this updating we can make AB-AODV more effective and responsive as a quick adaptation to conclude a number of mobile agents according to the changes in topology. In AB-AODV, to recognize the number of agents we have concluded manually the value of a number of agents. This makes AB-AODV more powerful and delay would also be minimized as manually configured work to evaluate mobile agents would be reduced.