The primary challenge of any directional mobile network is maintaining the links to the discovered neighbors for routing protocols to work efficiently. With higher mobility networks, such as fully directional UAV(Unmanned Aerial Vehicle) networks, the need for maintenance of links to the discovered neighbors through tracking becomes essential. In this thesis we propose a mobility prediction based polling mechanism for tracking the discovered neighbors to avoid neighbor rediscovery latency and reduce end-to-end delay of the entire network. First, we discuss various directional neighbor discovery schemes used in static and mobile directional networks. Second, we study different UAV-based mobility models and compare the performance of different UAV mobility models on the directional neighbor discovery process using various metrics. Next, we briefly look into different mobility prediction algorithms and select an algorithm to be employed in the proposed scheme. Lastly, we discuss a MAC (Media Access Control) integrated neighbor discovery scheme, PMAC (Polling-based MAC ) protocol, that incorporates polling for tracking of the neighbors. Having looked at the limitations of this scheme, we finally propose a mobility prediction-based polling mechanism, Smart-PMAC, for efficient and prolonged tracking of the discovered neighbors. We then compare the performance of the proposed scheme to PMAC using different performance metrics and prove that the proposed scheme outperforms PMAC in terms of tracking accuracy, average link lifetime, average neighbor links maintained, and packet delivery ratio.