A Regression based approach for link residual time prediction in MANETs
4 years ago
Creative Commons CC BY 4.0
Mobile ad-hoc network (MANET) is a collection of mobile terminals forming an infrastructure less and quick deployable network, which can communicate to each other via multiple hops or single hop. Such ad-hoc networks have always been important for various applications like defence applications especially for countries like India having boundaries and regions with large geographical diversity. Mobility attribute is a notable one in MANETs, as this leads to frequent topology changes which are the primary cause of route failure. A route is an ordered set of links, hence for predicting future availability of any particular route, it is important to estimate the availability of its currently available constituent links. This paper explores various link availability prediction model and proposes a least square polynomial regression-based statistical approach to predict the availability of link. Proposed approach assumes that movement of nodes are based on column mobility model i.e each node in the network is linearly moving with constant speed. Each node in the network periodically broadcasts hello packets to its neighbours to inform it’s availability in the network. Neighbour node receives hello packet and uses its signal strength to estimate distance between sender and receiver of hello packet. A monotonically decreasing signal strength of hello packets at receiver node indicates that nodes are moving away from each other and link between them may break in future so it starts link residual time prediction algorithm to predict the time when the distance between them will exceed the pre-defined threshold value. The proposed algorithm is simulated using NS 2.35. The performance of the algorithm has been analyzed for identified parameters. The results are also been compared by simulating other existing link prediction approaches based on interpolation.