Research On Key Issues Of Message Ferry-Based Routing In DTN

In recent years,Delay/Disruption Tolerant Network (DTN) has drawn more and more attentions, and gradually become a research hotspot. Due to the sparse distribution of the DTN nodes,the randomness of node mobility and the uncertainty of communication,the network topology frequently changes,communication links are often interrupted and the propagation delay is relatively long, which directly affect the DTN network performance. The traditional DTN network routing is using a store– carry– forward method. In this way,the messages are transferred through the opportunistic meeting between the network nodes. However,node mobility is not controllable; neither do the accessibility and the message transmission delay,which caused significant impact on DTN network performance. So people proposed a message Ferry-based routing mechanism to overcome the original routing problem. This mechanism provides an effective means through dynamic control of the Ferry nodes can effectively improve the performance of DTN network.This paper discussed the features of the message Ferry-based DTN routing and analyzed the advantages and disadvantages of the existing message Ferry-based routing algorithm firstly,and then concentrated to the target node selection and movement forecasting technology of Ferry and proposed new solutions. This work is mainly reflected in the following aspects:Firstly,we comparatively analyzed of the features of existing DTN network simulators,after which we designed and implemented a Ferry simulation platform based on ONE (Opportunistic Networking Environment). Our simulation platform can easily open or close the Ferry function of the nodes by the setting file and you can configure the routing pattern of the ferries.Secondly,we proposed the Multi-Constraint Target Selection Algorithm (MCTSA). MCTSA considers the distance of the target node,the message size and message generation time as constraints of the objective function,each constraint is given a appropriate weight parameter, which avoids the one-sidedness. The proportion of three parameters was optimized through simulation. Experimental results show that the message delivery rate,propagation delay and the overhead ration of the DTN have been greatly improved using MCTSA than the original single constraint target selection algorithm.Thirdly,we studied the movement pattern of the Ferry and proposed the location prediction method,which can meet the node earlier. Then we considered the specific implementation of the method in practice,proposing the Dichotomy-Based Meet Position Calculation (DBMPC) method. We evaluated the impact of DBMPC on the network performance using the Ferry simulation platform we designed. Experimental results show that the mobility location prediction algorithm brings stability enhancement of the network performance.