With the growth of wireless technology, Mobile Ad-hoc Network (MANET) has become popular and attracted ample research attention in recent years. MANET is very much useful in areas with limited or no existing communication infrastructures. The network is usually formed by mobile nodes using wireless communication. It is decentralized in nature and its topology changes very rapidly. Further, MANET has low node density, encounters interference in communication and consists of resource constraint nodes which result in frequent network disconnection. The dynamic nature of the network forces one to devise routing protocols that depend on the state of the network. Consequently, message routing is one of the major problems in MANET. Although several protocols depending on routing strategy and network structure have been proposed in the published literature, reliable message routing in MANET remains a fertile area of research. The paper, “E-B-M-Based Decision for Forwarding the Data in Intermittently Connected MANETs”, by D Jyothi Preshiya and C D Suriyakala, proposes an approach to message routing which takes the best forwarding decision based on the past encounter history and the behavior of the nodes. In their approach, the authors have considered parameters such as delivery certainty, ageing and transitivity, availability of the buffer space in the node and forwarding behavior of the nodes for choosing the best next node for forwarding the message. They claim that the approach minimizes resource consumption, reduces delivery delay and improves delivery ratio.

In many real-world data, the issue of class distribution is more pronounced due to the importance of minority class in the application at hand. For example, in financial fraud detection, thousands of transactions are genuine and legitimate, while a few are fraudulent transactions. While identifying such fraudulent transactions is important, it is also a difficult task. There is increasing interest in applying machine learning techniques to imbalance data in order to classify it effectively. As such, predictive accuracy, which is a popular choice for evaluating the performance of a classifier, fails to work while mining imbalanced datasets. The paper, “Classification of Skewed Data: A Comparative Analysis of the Performance of Select Classifiers”, by Banyabaishali Mohanty and Subhendu Kumar Pani, presents a study of classification accuracy of selected classifiers on skewed data using oversampling. It finds that Multilayer Perceptron, a type of neural network, performs better than Naïve Bayes and Decision Tree.

Cloud computing is the new buzz in software industry. It is a type of computing that relies on sharing computing resources rather than having own resources for one’s computing needs. The paper, “Virtual Appliances-Based Framework for Regulatory Compliances in Cloud Data Centers”, by Jitendra Singh and Vikas Kumar, suggests virtual appliances-based solution for regulatory requirements in order to ensure data security and privacy in cloud computing.

In spite of several measures, cyber crimes are increasing in number and sophistication, making the field attractive for research study. The paper, “Cyber Crimes: Evolution, Detection and Future Challenges”, by Raksha Chouhan, provides an account of this growing phenomenon.

Wireless sensor networks are becoming more and more popular due to various potential applications it promises. The paper, “Wireless Sensor Networks: Applications and Impact”, by Monika Kohli and Rohit Tiwari, presents a refreshing introduction to this growing field.

-- A C Ojha
Consulting Editor