AJMI Vol.07 Issue-02, (July - December, 2015) ISSN: 2394-9309 Aryabhatta Journal of Mathematics and Informatics (Impact Factor- 4.1) Double-Blind Peer Reviewed Refereed Open Access International e-Journal - Included in the International Serial Directories Aryabhatta Journal of Mathematics and Informatics http://www.ijmr.net.in email id- irjmss@gmail.com Page 9 Link Prediction in Mobile Ad hoc Network Prashant Singh * and D.K. Lobiyal # #Department of Information Technology, Northern India Engineering College, New Delhi #SC & SS, Jawaharlal Nehru University, New Delhi Abstract Mobility is the main cause of frequent route breaks in mobile ad hoc network. This results in frequent route changes that adversely affect the QoS in the network. QoS can be improved by using alternative routes before current route gets broken. Therefore, it is important to know how long a route will be available. The status of the routes can be determined if the availability of links between the nodes can be predicted. In this paper, we introduce an analytical model for link prediction. Since, the exact distribution of link availability is unknown, we have applied Laplace distribution. Epoch length and cumulative contribution of other factors causing link breaks between nodes have been considered as two key parameters of the distribution. Keywords – QoS, Link Prediction, Laplace Distribution, 1. Introduction Mobile Ad hoc Network (MANET) [1] is co-operative collection of mobile nodes communicating with each other through wireless links, without requiring any supporting infrastructure. In MANET [2] all the nodes can connect dynamically in an arbitrary manner due to their mobility. The nodes in the network behave both as autonomous nodes and as routers. In MANET, node mobility affects the quality of service requirement of applications by causing frequent link failures. Quality of Service can be assured by achieving a more deterministic network behavior [3, 4]. In MANET, because of the random mobility of the nodes network behavior is not deterministic mostly. Therefore, probabilistic approach can help in predicting the network behavior and thus providing QoS. In this paper, a probabilistic approach is used that can help in using the alternative routes before the current route is broken. This will help in better transmission and lesser end-to-end delay of message delivery. Availability of a route in future mainly depends on the availability of links between the nodes forming the route. Therefore, it is important to predict the future availability of a link that is currently available. Here, we have introduced an analytical model for link prediction using Laplace distribution. Since the exact distribution of link availability is unknown, we have applied Laplace distribution considering epoch length and cumulative contribution of other factors (such as energy dissipation, congestion, fading, etc.) as two parameters of the distribution.