Please use this identifier to cite or link to this item:
|Scalable Congestion Aware Replication framework for Mobile Ad hoc Networks
|Fault tolerance;replication;congestion detection
|Mobile Ad hoc Networks (MANETs) are types of decentralized networks which does not depend upon any existing infrastructure such as access points, routers or gateway. The nodes can communicate with each other by forming the network at fly, i.e., at any instant these can communicate with IEEE 802.11 a/b/g standards. The network consists of mobile nodes that are multi-hop, self-configured connected by wireless links. Each node randomly moves and changes its positions arbitrarily. So, the topology of network changes very frequently. In MANETs, dynamic mobile nodes are stationed in such a manner that communication between nodes does not rely on any existing network infrastructure. Due to high velocity of nodes in MANETs, fault can occur at any time due to the instability of the links. This thesis provides solution to overcome the difficulty of faults that occur in the network by tolerating fault and providing data availability. The solution proposes for fault tolerance is by using the technique of replication of data items at the backbone nodes. A new Scalable Congestion Aware Replication (SCAR) framework is designed that overcomes the fault and increases the data availability. In this the data is replicated over the backbone nodes that are created in the network by the marking and pruning process. Thus in case the data over one node is not present it can get that from the nearby backbone node. The data is replicated depending upon the cost function of that particular data item. The data item having more value of the cost function that will be replicated in comparison to the item having less cost function. The designed algorithm is adaptive with respect to the rate of replication, time to overcome the fault and success ratio.
|Appears in Collections:
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.