Informatics Publishing Limited

T. Mekala ¹, P. Nandhini ²

Affiliations
1 CSE Dept., M. Kumarasamy College of Engineering-Karur, IN
2 Cognizant Technology Solutions, Tamilnadu, IN

Abstract

Clustering is the task of finding natural partitioning within a data set such that data items within the same group are more similar than those within different groups. At present all available clustering techniques for web usage mining are based upon usage patterns derived from user's page preferences. To improve the efficiency of usage pattern, the similarities in user's access behavior based on time locality of user's navigational acts are used. The tuning algorithm defines cluster that show similar visiting behavior at the same time period and users page preferences by varying the priority given to the time visit.

Keywords

Web Mining, Web Users Clustering, Navigation, Access Time.

Full Text

   

N. Mahendran ¹, T. Mekala ²

Affiliations
1 Department of Electronics and Communication Engineering, M. Kumarasamy College of Engineering, Thalavapalayam − 639113, Karur, Tamil Nadu, IN
2 Department of Computer Science and Engineering, M. Kumarasamy College of Engineering, Thalavapalayam − 639113, Karur, Tamil Nadu, IN

Abstract

Objective: To minimize the energy frenzied around cloud datacenters. The proposed system provides a dynamic and continuous load prediction along with existing heuristic load prediction to improve the adaptation performance for green computing through Markov chaining of resources. Methodology: The trouble of energy competence through resource management for a large-scale cloud environment to facilitate hosts sites is resolute here. In apply there will be number of underutilized servers in the datacenter which guzzle bulky part of established energy. To minimize the energy frenzied around datacenters the existing system outlines a distributed middleware architecture and presenting one of its key elements, a gossip protocol, Skewness algorithm and load balancing that meets our design goals: justice of resource allocation with respect to hosted sites through overload escaping and efficient edition to load changes and scalability in terms of together the numeral of machines and sites thereby turning off the unused servers for green computing through Markov Chaining Model. This model is scalable enough to signify systems composed of thousands of resources and it makes potential to represent both physical and virtual wealth exploiting cloud explicit concepts such as the infrastructure elasticity. Results: To certify the model, simulation is conducted within the Network Simulator (NS) 2.28 have platform with GCC 4.3 and Fedora 13. The load prediction algorithm and Markov chain model achieves both overload avoidance and green computing for systems with multiresource limitations. Application: Green Wireless Network: To improve the energy efficient, data retrieval and resource service based on coaching, computing and networking in the area of Green Wireless. Green Big Data: To reduce the power consumption, big data merge the concept with green computing. Green job Scheduling: Each Server has different jobs, to save the energy server merged with green computing. Green Cloud data center: To use the green computing in cloud data center to increases the energy efficiency.

Keywords

Green Computing, Load Prediction, Markov Chaining, Virtualization

Full Text