Abstract Internet of Things (IoT) plays a vital role in Next Generation Networks. Ample number of research works in IoT is carried out in developing countries like India. Research and Development units of industries are working on connecting tiny devices and objects to infer and to measure environmental and ecological resources. Domestic applications are also in line with this research. This... Internet of Things (IoT) plays a vital role in Next Generation Networks. Ample number of research works in IoT is carried out in developing countries like India. Research and Development units of industries are working on connecting tiny devices and objects to infer and to measure environmental and ecological resources. Domestic applications are also in line with this research. This paper proposes an integrated smart environment based on IoT. Several sectors like agriculture, security and emergency, banking, Surveillances, meteorology, health care, education, government – e services, domestic appliances monitoring, traffic surveillance are integrated and the various objects and devices are connected using RFID technology. This paper also deals with how various sectors are connected by means of RFID technology and sensor networks. It also brings forth an idea of establishing IoT information Kendra which infers and processes the data extracted by the various sectors in the smart environment. Functioning of a cloud application based computing and data center and an administration and management center attached to IoT information Kendra is also discussed. Read More Read Less

Abstract Case-based reasoning has been widely adopted for decision support. A major operation in the CBR is the retrieval of similar cases. However, this operation has some weaknesses. One among them is the retrieval of several similar cases. On the other hand, sequentially processing all cases with a similarity has a complexity in presence of a great case base and several features.... Case-based reasoning has been widely adopted for decision support. A major operation in the CBR is the retrieval of similar cases. However, this operation has some weaknesses. One among them is the retrieval of several similar cases. On the other hand, sequentially processing all cases with a similarity has a complexity in presence of a great case base and several features. So, improving retrieval has been focused by a considerable amount studies using sequential calculation, non-sequential indexing, classification algorithms and Nearest Neighbor matching, etc., while others use hybridization of CBR with other reasoning methodologies. In this paper, we propose a novel approach based on CBR and clustering to improve the retrieval operation, and impacts positively the whole reasoning process. Our aim is to propose an available strategy for the retrieval task and also a valid decision support model. Finally, we present preliminary results and suggestions to extend our approach. Read More Read Less

Abstract In recent years, software fault classification became an important research area for the development of reliable and high quality software products. The fault classification helps in identification of fault software modules and allows the developers to concentrate on that module. It helps the developer to save the time and control financial losses to industry. Therefore, in this paper, a novel... In recent years, software fault classification became an important research area for the development of reliable and high quality software products. The fault classification helps in identification of fault software modules and allows the developers to concentrate on that module. It helps the developer to save the time and control financial losses to industry. Therefore, in this paper, a novel fault classification method based on feature ranking algorithms and ensemble techniques is proposed. The number of features available in the metrics are selected to represent the fault using feature ranking algorithms and operated on ensemble techniques to check the performance. Also, various hyper parameters are tuned for the ensemble techniques to identify the best model. The experimental result demonstrates the good results for bagging with K-nearest neighbor and random forest in comparison with other methods. Read More Read Less