Fault Tolerant Strategies in MANET

Author Name(s): Ms. Fahmina Taranum , Mr. Khaleel Ur Rahman Khan
Author Email: ftaranum@mjcollege.ac.in1

Abstract

A fault tolerant system for Mobile Adhoc Networks (MANET’s) is a system, which works well even in case of failures. The aim of the paper is to handle the issues like network partitioning, cluster head management and data transmission. The parameters like mobility speed, battery value and reach ability range are used for effective implementation. The communication is done via cluster heads with highest mobility speed. Replicas of cluster heads are created for the non-reachable nodes. The number of replicas is equal to double the value of nonreachable nodes. The random values of the replicas for non-reachable cluster head node are better when compared to the old values. The data is send from the sender to the destination node through the cluster heads. The nodes with the three maximum mobility speeds are selected for communication. The data moves from the third maximum mobility speed cluster head node to second and then to the first.

Key words

MANETS, TTL, Battery value, Mobility range, Reliability factor.

Introduction

Mobile ad-hoc networks (MANET’s) composed of mobile nodes which can migrate while communicating with each other thereby resulting in partitioning of network. If the node moves within the scope then there will be no network partition. In case of network partitioning there are two cases to consider, one in which the mobile node is a cluster head or simply a communication node. If an unavailable node is cluster head then a new cluster head needs to be elected from the available nodes and when the old cluster head comes back to region of reach ability either it should take over the charge or the new cluster head would retain its charge. This decision has to be taken by the members of the cluster. In the second case if the node is one amongst the node with the three maximum speeds then a replica is created else it would be ignored. Another challenge is to overcome network partitioning. Network partitioning is the condition that exists after a network connection between any two groups of systems fail or a nodes moves out of defined reach ability. When this happen, systems on both side of the apportion can restart their applications resulting in duplicate services. The most serious problem caused by a network
partition is that it affects the data on shared storage. Replication is employed as an approach to overcome this problem and to ensure higher service. Replication is widely used for faulttolerance, scalability and performance. The failure of one replica does not stop the system from working as available replicas can take over the tasks of the failed replica. Finally, replication can provide fast local access, even if clients are geographically distributed. Threshold is applied to parameters to declare a node as selfish. Selfish nodes are the nodes participating in the network which hesitate to forward the packets in order to save their resources like low energy constraints.

Conclusion

The proposed approach works well in case of failure of cluster head as the data is passed through the available replica of the cluster head. The system behaves efficiently as the data passes through cluster head always to other cluster head till it reaches the destination node. This application makes a fault tolerant system as it works properly in the presence of a failure node. In this application, clustering is performed where in certain number of nodes are divided into a cluster. Each cluster has a cluster head through which data communication is done. Leader election algorithm is used to elect a cluster head. Election of leaders must use the highest remaining battery life and the lowest mobility range in the election process. Leader election is done in the presence of some selfish nodes, unavailable nodes and nodes which are outside the network. The strategies can be summarized as follows: 1. Solves the network partitioning issue using replication technique 2. The efficient leader is elected in the presence of selfish nodes 3. Maintains replica of the failed node 4. Maintains coordination among the replicas even if the leader node is dead

865 total views, no views today

Download PDF File

About the author: admin