A Network Layer Service Discovery Approach for Mobile Ad hoc Network Using Association Rules Mining

Mobile Ad hoc Networks (MANET) are infrastructure less networks that are formed on the fly between groups of mobile nodes. Due to the vigorous nature of such networks, service discovery in MANET is a very cumbersome job. In this paper, we have proposed an integrated approach to service discovery in MANET that works at network layer to discover services that also piggy-backs responses of possible future requests with current response. Due to this integrated approach, the scheme provides an appropriate route to access the service along with discovering a service provider. The proposed approach is based on a proactive component that periodically broadcasts service advertisements and a reactive component that discovers a service on-demand when a service request is not satisfied locally. To discover the services, the reactive component relies on service oriented routing algorithms to propagate services on the network. In this paper, we have discussed two service oriented routing algorithms. The first algorithm is based on Ad hoc On-Demand Distance Vector Routing and is adopted from literature. The second algorithm is based on Dynamic Source Routing protocol and has been proposed by us in this paper. In order to improve efficiency further, our approach predicts future service requests and appends responses of these requests with current response. In order to achieve this objective, the proposed approach applies Association Rules Mining on the service requests history to discover correlation among services. The mining results are then used to predict future service discovery requests and then attaching future requests information with current response. The proposed scheme has been implemented on JIST/SWANS simulator. The results of the simulation show that the proposed approach is robust, efficient and scalable.