Research On The Buffer Management In Delay-Tolerant Networks Based NS-3

Deep space communications become the focus of the development of science and technology.Its communication features such as long time delay, indirect connection and high bit error rate, make the existing TCP/IP protocol is no longer applicable. Delay tolerant network has been proposed, which gives some solutions to the challenge network.Delay tolerant networks(DTN) have is based on storage-forwarding mechanism. Each network node must have persistent cache capabilities. Good buffer management strategy with the routing algorithm can effectively control congestion and improve network performance. Now there are a lot of researchers researching the DTN network, and they put forward a lot routing strategies but not an effective congestion control strategy has been put forward. The cache space is an important resource in the network, so the research on the cache is promising prospects.In this paper, the first part introduces the basic principles of deep space network and the DTN network architecture and related research. We summarize the existing DTN network routing strategy and congestion control strategy. In order to improve the delivery rate of the message, the paper improves the buffer based congestion algorithm proposed by J.Lakkakorpi, which can efficiently increase the delivery ratio and the utilization of cache. This mechanism is a node actively transmits its buffer occupancy and the bundle information to the adjacent nodes. The adjacent node can obtain the congestion of each node in the small area, and send the messages to the node which has much buffer space.The source node no longer need to be aware of whether there are available paths. In addition, to obtain the topology of the entire network, the source node will receive a lot of feedback so that the network resources are not fully utility.This paper illustrates the process of software design in detail.Finally, the DTN simulation model is set up under NS-3 simulation platform. We performed simulations for real-world mobility traces the San Francisco taxi mobile model. The simulation in this paper possesses 30 nodes and 116 nodes, the simulation duration is 3600 s. Through the delivery ratio, hop count, buffer occupancy, delays and so on, the two algorithms are compared and we verified the effectiveness of our method and the performance of the network. Concluded the original buffer congestion control algorithm is not suitable for low visibility network which is very common, especially in the deep space network. The improved algorithm can be applied to the network, and it shows a good characteristic.