Search

One of the challenging security threats to wireless mesh network (WMN) is their vulnerability to routing protocols. This vulnerability is mainly caused by certain internal attacks, one of which is known as a wormhole attack. This attack is launched by two or more malicious nodes which gives a false illusion of shortest path through them via a tunnel. They formed a tunnel via encapsulation, which restricts the increment of hop count during the traversal through intermediate nodes and therefore launches a wormhole attack between source and destination. In this paper, we proposed a detection mechanism based on the calculation of round trip time (RTT) and processing time to identify the malicious nodes forming wormhole attack. Our proposed work prevents the AODV routing protocol against the wormhole attack in WMNs. The simulation of our proposed work had done using NS-3 simulator, and the results show that the performance of our detection algorithm improves over the existing detection techniques against wormhole attack.

Wireless mesh network (WMN) is considered as one of most emerging field for the research work in today's networking area due to its various approaches available to handle the network resources. Various architectures have been proposed so far and main key challenge for the design of this network is their vulnerability to security attacks. Therefore an efficient key management is required to overcome from such security exploitation such as unauthorised access to the network. WMNs need to be clustered into multiple groups to achieve fast and efficient security mechanism to overcome from security attacks. Therefore, we have proposed a group key agreement protocol based on Diffie-Hellman protocol and digital signature to authenticate the mesh entities within the group. The authentication took place between trusted authority (TA) and mesh clients with efficient communication and computational cost and results in reduction of packet loss by an attacker in wireless network which improves the overall performance of the WMNs. Our experimental result proves that it achieves better communication and computational cost with non-repudiation attack.

In Natural Language Processing (NLP), Parts of Speech (PoS) tagging is one of the most crucial steps of pre-processing in every language system. PoS tagging is the process of selecting the best suitable ‘part of speech category’ or ‘lexical class label’ for each token in a phrase in a natural language. It is usually the initial stage of an NLP task like machine translation, with further stages, including chunking, parsing, etc. The key objective of our work is to create an English-to-Mizo Machine Translation (MT) system using PoS tag corpus. Here, we use factored Phrase-Based Statistical MT (F-SMT) to address the issue of text translation from English into Mizo, one of the under-resource language pairs. During the process, we discovered that better accuracy of the system can be achieved by combining the training with PoS-featured data. Experimental results achieved by employing automatic evaluation metrics demonstrated that our F-SMT with PoS tagging model outperformed the baseline phrase-based model and other factored models when applying the various n-gram parameters. Our F-SMT with PoS tagging exhibits an increase in the automated scores of translation; it achieved scores of 14.27 (BLEU), 47.70 (F-measure), 30.30 (METEOR), 37.60 (Precision), 52.40 (Recall), and 75.30 (TER).