利用巡迴式中繼結點在耐延遲網路下進行路由。;Delay-tolerant networks (DTNs) represent mobile wireless networks that are generally characterized by no end-to-end paths from a source to a destination; examples of such networks include military networks, sensor networks, wildlife tracking networks, and vehicular ad hoc networks. In such sparse networks, nodes must carry messages until they contact with appropriate nodes before forwarding the messages. This process is denoted as a store-carry-and-forward routing scheme.
Although routing algorithms in DTNs are being increasingly designed, the network performance associated with such algorithms is still limited because of node mobility limitations. Therefore, a message ferry scheme was proposed for improving network performance. A ferry is a specific node that provides controllable movement and nonrandomized contact opportunities in DTNs. In conventional ferry-assisted DTNs, a ferry typically moves repeatedly along a predefined route comprising several hotspots. However, if the network environment changes (specifically, if such hotspots no longer exist), such a movement type may fail because ferry movement or ferry route definitely influences the network performance.
This paper proposes a geographic ferry movement (GFM) scheme involving determining hotspots and designing ferry movement patterns. The GFM scheme entails using node contact history to determine hotspots dynamically. Moreover, we formulated hotspot travel scheduling problem as a Hamiltonian path problem and used classical solution in graph theory to design a route. We used the branch-and-bound algorithm and the nearest neighbor algorithm to determine a trade-off between the length of a hotspot tour and network performance.
Finally, we observed the difference in the performance of three routing protocols according to four mobility patterns before and after 1–10 ferries joined the networks. The results revealed that the GFM scheme can be used to effectively determine hotspots and that ferries can clearly improve network performance considerably.
