The ad hoc network is defined by the mobile nature of the nodes and the removal of the requirement for an infrastructure based network, i.e. this mobility and wireless nature is one of the main features of the ad hoc networks and helps it to in any way therefore to be used off site. The lifetime of an ad hoc network can be defined as the time when the network starts working until one of the nodes drops out of the network due to the battery being drained.
When clusters are formed in the network, one of the nodes becomes the cluster head (there may also be multiple cluster heads depending on the selection parameters), which bears the responsibility of maintaining communication with all other nodes. This responsibility of maintaining communication links drains the battery very quickly as this node has to make more number of communications compared to a normal node and the node goes out of the network. Also the Cluster Head consumes the maximum of its battery compared to the rest of the nodes.
So, if the number of nodes under one Cluster Head is more compared to the rest of the Cluster Heads, then this node will drop out of the network prematurely. Hence the energy consumed to communicate with the different nodes in the networks, forming the cluster, checking for live nodes, etc. In the third chapter we did the literature review of the Ad Hoc networks and its components.
The improvements made with the base algorithms implement the balanced distribution of the nodes among all the ClusterHeads allowing each ClusterHead to have almost the same load on it.
LITERATURE REVIEW
- Wireless Network
- Ad Hoc Networks
- MANET
- Clusters
A wireless ad hoc network is a decentralized wireless network where the network does not depend on pre-existing infrastructure, such as routers in wired networks or access points (APs) in managed (infrastructure) wireless networks. The earliest wireless ad hoc networks were "packet radio" networks (PRNETs) from the 1970s, according to the DARPA-sponsored ALOHA net project. This decentralized nature of ad hoc networks makes them suitable in places where a single central node acting as a base station does not work efficiently or the terrain is not suitable for the placement of fixed nodes that are connected via wired links.
MANET - A mobile ad hoc network (MANET) [18], sometimes known as a mobile complex network, is a self-configuring network of mobile devices connected by wireless links. A MANET [22], also known as a mobile complex network, is a self-configurable collection of mobile devices that connect via wireless links. Each device in a MANET is connected to another through wireless links, but since the network is an ad hoc network, a node can move in any direction without any restrictions.
VANET - A vehicle ad hoc network (MANET) is used to communicate between vehicles and/or road equipment. Intelligent VANET - A vehicle ad hoc network (InVANET) is used to communicate between vehicles and/or road equipment and to act intelligently in case of collision, minor hitches, etc. High Availability Clusters - Also known as Failover Clusters [24], are implemented primarily with the aim of improving the availability of services provided by the cluster.
For example, a single computing task may require frequent communication between nodes—this implies that the cluster shares a dedicated network, is densely located, and likely has homogeneous nodes. Grids serve to manage the allocation of jobs to computers that will perform the work independently of the rest of the grid cluster. However, it does extra work with respect to regular stations, and therefore can become the bottleneck of the cluster.
In each cluster, one node is elected to be the cluster head while the rest of the nodes become common nodes. ClusterHead coordinates intra-cluster transmissions, handles inter-cluster traffic, and delivers all packets destined for the cluster; it can also exchange data with nodes that act as gateways to wired networks. In cluster-based network architectures, the lifetime of the network is strongly related to the failure of the Cluster Head.
This assignment of nodes to a ClusterHead is one of the most important determining factors for the life of a network. Even after the allocation of this, the balancing of load in terms of the number of nodes served by the ClusterHead is important.
CLUSTERING ALGORITHMS
- Related Work
- Parameter Identification
- Highest Connectivity Algorithm
- Lowest ID Algorithm
While implementing all these algorithms, we have used a set of common algorithms to calculate the distance between the nodes, random work deduction and the mobility of the nodes. The highest connectivity algorithm [6] [7] selects a cluster head based on the degree of connectivity with the other node. So when the algorithm starts executing, the node with the highest connectivity with the other nodes becomes the cluster head.
From the remaining set of nodes, the next group head is selected using the same conditions/parameters. The HC algorithm helps to identify the cluster head very quickly, but we have a serious problem of one of the nodes being loaded with more responsibility compared to the other nodes. Once this is done, we subtract the random work done by the cluster head to communicate with the rest of the nodes.
A major flaw of this election algorithm is that the number of nodes under one cluster head is relatively high compared to others. This effect is also exacerbated when increasing the number of nodes in the sample area. This algorithm picks up the node with the lowest ID and checks the connectivity with the neighboring nodes, and with the help of these nodes it forms the cluster.
So when the algorithm starts executing, the node with the lowest ID becomes the ClusterHead and all the nodes within communication range are covered under this ClusterHead. Then the next node with the lowest ID is selected to become the ClusterHead, with the same conditions/parameters selected. This algorithm is a bit better than the HC algorithm because it allows different nodes to become the ClusterHead.
But this algorithm has a flaw; the node with PID 1 will always become the cluster head, will drain its battery faster and will be the first to drop out of the network most of the times.
PROPOSED ENHANCEMENT
Proposed Enhancement
The choice of the ClusterHead is based on a Fuzzy decision taken by the nodes that will come under different ClusterHead. For this fuzzy selection we use a few more parameters, namely work budget left and number of nodes under the ClusterHead. NC: Set of distinct ClusterHead that can take the current node under its own cluster.
In the above algorithm, first calculate the average work budget remaining for all the ClusterHead and the average number of nodes under a ClusterHead. We then pick up the ClusterHead with the maximum work budget remaining which is the maximum battery life remaining. Similarly, we select the ClusterHead with the least number of nodes below it and store it under the value T2.
Here we use the MOD function as we want the difference between the values and not the actual value. Now if you select the minimum value from this array, we get the desired ClusterHead for the current node. We multiply because if WB is high but the difference between NN and AVG (NN) is low, we get a lower value compared to the rest of the combinations.
The ClusterHead, which has the lowest value, will have a relatively higher work budget and a smaller number of nodes will be selected under it. This part of the improvements is made to reduce the effect of increasing the load when a new node. When a new request arrives at a ClusterHead, the ClusterHead searches for all ClusterHeads that can take this node under its cluster.
SIMULATION AND RESULTS
Simulation Bed
Results
The network initially shows low network lifetime in LID because the number of nodes is less.
CONCLUSION
REFFERENCES
