A. Unnikrishnan

A PROBABILISTIC APPROACH TO REDUCE THE ROUTE ESTABLISHMENT OVERHEAD IN AODV ALGORITHM FOR MANET

Mobile Ad-hoc Networks (MANETS) is a collection of wireless nodes without any infrastructure support. The nodes in MANET can act as either router or source and the control of the network is distributed among nodes. The nodes in MANETS are highly mobile and it maintains dynamic interconnection between those mobile nodes. MANTEs have been considered as isolated stand-alone network. This can turn the dream of networking “at any time and at any where” into reality. The main purpose of this paper is to study the issues in route discovery process in AODV protocol for MANET. Flooding of route request message imposes major concern in route establishment. This paper suggests a new approach to reduce the routing overhead during the route discovery phase. By considering the previous behaviour of the network, the new protocol reduces the unwanted searches during route establishment process.

Recursive Decision Tree Induction Based on Homogeneousness for Data Clustering

Data mining is an analytic process designed to explore data in search of consistent patterns or systematic relationships between variables. To build a model for data mining, both supervised and unsupervised learning techniques are used. In this paper we try to make use of a supervised learning technique called classification tree commonly called decision tree to cluster the similar featured attributes of large datasets. The algorithm takes an image of plotted data values as the input and inducts a decision tree accordingly. The decision factor to form the tree is a measure of homogeneousness of the data pixels in the region. Reverse merging of leaf nodes are done to make clusters based on their domain density.

Information Content Extraction on Quad Trees for Active Spatial Image Clustering

Spatial Mining differs from regular data mining in parallel with the difference in spatial and non-spatial data. The attributes of a spatial object is influenced by the attributes of the spatial object and moreover by the spatial location. In this paper, we propose a new algorithm for spatial mining by applying an image extraction method on hierarchical Quad tree spatial data structure. Homogeneity of the grid is the entropy measure which decides the further subdivision of the quadrant. Finally, the algorithm proceeds by applying low level image extraction on domain dense nodes of the quad tree.

Visualization of Mixed Attributed High-Dimensional Dataset Using Singular Value Decomposition

The ability to present data or information in a pictorial format makes data visualization, one of the major requirement in all data mining efforts. A thorough study of techniques, which presents visualization, it was observed that many of the described techniques are dependent on data and the visualization needs support specific to domain. On contrary, the methods based on Eigen decomposition, for elements in a higher dimensional space give meaningful depiction. The illustration of the mixed attribute data and categorical data finally signifies the data set a point in higher dimensional space, the methods of singular value decomposition were applied for demonstration in reduced dimensions (2 and 3). The data set is then projected to lower dimensions, using the prominent singular values. The proposed methods are tested with datasets from UCI Repository and compared.

Performance improvement of multiple Connections in AODV with the concern of Node bandwidth

Mobile Ad-hoc Networks (MANETS) consists of a collection of mobile nodes without having a central coordination. In MANET, node mobility and dynamic topology play an important role in the perform ance. MANET provide a solution for network connection at anywhere and at any time. The major features of MANET are quick set up, self organization and self maintenance. Routing is a major challenge in MANET due to it’s dynamic topology and high mobility. Several routing algorithms have been developed for routing. This paper studies the AODV protocol and how AODV is performed under multiple connections in the network. Several issues have been identified. The bandwidth is recognized as the prominent factor reducing the performance of the network. This paper gives an improvement of normal AODV for simultaneous multiple connections under the consideration of bandwidth of node.

Enhanced Spatial Mining Algorithm Using Fuzzy Quadtrees

Spatial Mining differs from regular data mining in parallel with the difference in spatial and non-spatial data. The attributes of a spatial object is influenced by the attributes of the spatial object and moreover by the spatial location. A new algorithm is proposed for spatial mining by applying an image extraction method on hierarchical Quad tree spatial data structure. Homogeneity of the grid is the entropy measure which decides the further subdivision of the quadrant. The decision for decomposition to further sub quadrants is based on fuzzy rules generated using the statistical measures mean and standard deviation of the region. Finally, the algorithm proceeds by applying low level image extraction on domain dense nodes of the quad tree.