Space Delay-Tolerant Networks Routing Using Artificial Neural Networks

Most of the present communication networks count on having an end-to-end association between the sender and receiver to be able to establish a connection; unfortunately, that is not always possible due to obstacles or environmental difficulties. Delay Tolerant Networks (DTN) is the type of networks that is specially designed to overcome those difficulties. Instead of having an end-to-end connection between nodes, In general, DTN routing protocols use a ”save when no link” and ”forward when possible” policy. Machine learning is an engineering field that uses statistical techniques and use it to make a system learn. This research will use machine learning technology, Artificial Neural Networks (ANN) specifically and deploy it to perform DTN routing. The neural network will be taught a database of previous patterns of network node Contact Graph (CG), in other words, the neural network will learn on when nodes will have direct sight of each other. After training the network, a network node should be able to make the best decisions about the best next hop to reach the destination with minimum acceptable error rate. The main advantage is that each node will be already trained on the best possible routes before the actual routing decisions need to be made, so by the time that any node is handed a bundle, it already knows where to forward the bundle instead of just flooding the network almost randomly trying to reach the destination like traditional routing protocols. A well-trained DTN node will be knowing and waiting for the next hop to appear in sight, and it will deliver its data as soon as the node-to-node connection is established. To the best of our knowledge, this is the first work that introduces and utilizes artificial neural networks to implement delay-tolerant networks routing.Engineering Technology, Department o