The Study and Design of Buffer Management in DTN

• Main Authors: Jhan-Hua Liou, 劉展華
• Other Authors: Shou-Chih Lo
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/02995283903697296332

碩士 === 國立東華大學 === 資訊工程學系 === 98 === Delay Tolerant Network (or Disruption-Tolerant Network, DTN) is a new type of routing concept for wireless communication applications in recent years, such as interplanetary networks, sensor networks, power-constrained wireless communication equipments, etc. They all belong to the scope of delay tolerant networks. DTN is different from the traditional Internet, since its sender and destination may not often exist in an end-to-end path. But in the past Internet framework, routing protocols must have a complete transmission path from the sender to the destination at any time. Consequently, it is necessary to use a new mechanism called store- carry-and-forward that is proposed to transmit data and solve the frequent interrupted connection problems for such special communication environments. When the connections are interrupted, the nodes will not discard data, but the data will be carried until they meet other suitable nodes, and then forward these data. Therefore, each node may be required to carry and store the messages in the buffer for a long time by itself. But in the real communication environment, buffer space is limited, so buffer management mechanism is a very important target affecting transmission efficiency. In this paper, the current routing mechanisms for DTN are introduced firstly. Moreover, we list the current common policies for buffer management in DTN and then propose the queuing policy that we want to compare. Besides, individual performance for each queuing policy that is observed in the simulation. By using a utility-driven approach, we try to find more suitable utility function for real environment in many affected buffer queuing policies. Finally, in the simulation we use the real trace files to evaluate our proposed mechanisms and compare them with other existing ones. In the simulation results, we show that our mechanisms have a superior performance on delivery ratio and throughput.